Around 60,000 compressed air systems are in operation in German companies. Together, they consume 16.6 terawatt hours every year, which corresponds to seven percent of the total electricity consumption of domestic industry. "The costs for this could be reduced by up to 30 percent," says Professor Sauer, Head of Resource Efficient Production at Fraunhofer IPA and Director of the Institute for Energy Efficiency in Production (EEP) at the University of Stuttgart.

For him, one of the greatest potential savings is the fact that the vast majority of compressed air systems have so far been inefficient. The reason: they are full of leaks. Holes and kinks in the hoses or leaking connectors: All this is difficult to detect. This is because often not all parts of a compressed air system are easily and safely accessible and the leaks are so tiny that they are very difficult or even impossible to see with the naked eye.

Until now, an ultrasonic measuring device has been used to detect the frequency ranges at which the air escapes, which are inaudible to humans. Most companies only make this effort once a year at most, or simply live with the leaks.

Holes, kinks and leaking connectors
Christian Dierolf and his colleague Christian Schneider want to use artificial intelligence to identify leaks in compressed air systems and put an end to waste. As a first step, the two researchers have set up a demonstration system. In it, the compressed air flows either through intact hoses or through hoses with barely visible holes, kinks and leaking connectors - the most common leaks in compressed air systems in industry.

Whichever path the compressed air takes, it makes no difference to the naked eye: the actuators do their job. However, the demonstrator measures whether the air is flowing through the hoses with more or less pressure, determines the flow rate, the position of the actuators, the status of the valves and records ultrasonic signals.

Demonstrator as a database
All of this is stored synchronously in a cloud. "The demonstrator therefore creates the basis for our data-driven production research, for example by training self-learning algorithms," explain the researchers. These algorithms will later be transferred to industrial applications. There, they will not only detect and localize leaks, but in future will also display the name and order number of the affected component via an app. The person responsible for the compressed air system will then no longer have to search the catalog for a long time. Instead, they can procure a replacement with just a few clicks and thus keep downtimes to a minimum.

"In addition to classifying leaks, the focus of the research is on identifying the actuators present in the machine's compressed air network with minimal measurement effort," says Dierolf. However, like many of the researchers' other ideas, this is still a long way off. The seminar "Intelligent compressed air - identifying potential and increasing efficiency with Industry 4.0 methods" on November 6, 2019 at the Fraunhofer IPA in Stuttgart will show which measures, including with the help of Industry 4.0, can already be used to exploit existing efficiency potential and significantly reduce costs. as