Predicting a motor’s future
Health check for electric motors. A new development makes it easier to predict remaining useful life and troubleshoot faults.
Predictive maintenance is a useful means of minimizing downtimes for small electric motors. And now it’s even easier, thanks to a new development from Siemens. A combination of the digital twin and artificial intelligence makes it possible to determine a motor’s remaining service life and identify effective ways to extend it.
Without small electric motors, industry comes to a standstill. Whether it’s pumps, blowers, or compressors, whether it’s cranes, conveyor belts, or packaging machines – they drive entire fleets of machines. The failure of one motor can quickly have costly consequences, including production losses, express repairs, and possible replacements.
That’s why companies are increasingly relying on “predictive maintenance” to detect vulnerabilities before they cause damage. As part of a service agreement, Siemens Digital Industries is offering the “Predictive Service Assistance” app for motors in the low-voltage range. In combination with intelligent analysis algorithms, the app supports predictive maintenance. Until now, industry has had no way to estimate remaining useful life (RUL), meaning no way to determine from current status values whether these values could lead to a total breakdown. “Every service technician can tell you what a great step forward this would be,” says Matthias Erlwein, manager in Customer Service at Siemens. “Even with decades of experience, they still can’t look into the future.”
Digital twin plus artificial intelligence
For the first time, Siemens researchers have now found a way to do exactly that and predict the remaining service life of low-voltage motors. “We use a digital twin that serves to identify the effects of various factors on remaining useful life,” says Vincent Malik, an expert in simulation and the digital twin. A digital twin is a virtual model of a machine, process, or production plant that contains all the data and simulation models relevant to its deployment.
To produce the digital doppelgänger of a low-voltage motor, Malik’s team performed several steps. First of all, they used high-fidelity simulations to develop a digital twin of a motor family. Based on this digital twin, they then used artificial intelligence with a running motor to determine additional relationships – for example, the exact way in which wear conditions affect sensor values. The result was a new, sophisticated twin. All Malik and his colleagues had to do was enter the data for an individual motor – such as size and weight – in the digital twin to obtain a detailed model.
Decide early whether there is need for action
The customized twin can identify causal chains for the potential failure of an individual motor – for example, pending bearing damage. "The older the bearings, the more they vibrate, which in turn increases motor vibration and material fatigue,” explains Christian Wolf Pozzo, who develops service for electric machines at Siemens Digital Industries. "Other possible causes may be mechanical factors or adverse ambient conditions. Within a matter of seconds its remaining useful life can be calculated." Based on that the Siemens service expert can decide whether there is need for action, e.g. to reduce power or to exchange certain components. Thus, product data become valuable knowledge to optimize processes effectively in order to extend lifetime, reduce costs and increase quality.
“This approach can just as easily be applied to other machines with rotating parts, such as compressors, blowers, and pumps,” says Malik
Hubertus Breuer - March 2020
Subscribe to our Newsletter
Stay up to date at all times: everything you need to know about electrification, automation, and digitalization.