Leaks are responsible for a large proportion of the losses in a compressed air system - on average, up to 30% of energy is 'blown away' via leaks. With 60,000 compressed air systems in Germany and an electricity consumption of 16.6 TWh per year, eliminating compressed air leaks alone could save as much electricity as Hamburg and Munich together consume each year.

For Prof. Sauer, Head of the 'Resource-efficient production' division at the Fraunhofer Institute for Manufacturing Engineering and Automation IPA and Director of the Institute for Energy Efficiency in Production (EEP) at the University of Stuttgart, this is reason enough to focus on compressed air leaks. Together with the researchers at Fraunhofer IPA, he wants to use AI to detect leaks in compressed air systems quickly and accurately. To this end, Fraunhofer IPA has worked with Mader to construct a compressed air demonstrator that provides the data basis for this. It is intended to create the basis for Fraunhofer IPA's data-driven production research, for example by training self-learning algorithms.

The leaky system

All measurement and sensor data is also automatically transferred to the Fraunhofer IPA's Industry 4.0 platform 'Virtual Fort Knox' via OPC UA and processed there for further analysis.

© Fraunhofer IPA / Mader

In order to understand how a leak in the system manifests itself, a handling system was first set up that simulates automated production. A maximum of four different scenarios can be selected at each of four stations - pressing, swiveling, vacuum pick-up and transporting: all tight, kink in the hose, hole in the hose and mechanical leakage. The measured values - volume flow and pressure - can be read off the display at any time.

Replacement for ultrasound?

All measurement and sensor data is also automatically transferred to Fraunhofer IPA's Industry 4.0 platform 'Virtual Fort Knox' via OPC UA and processed there for further analysis. However, since such IIoT projects rarely have a completely uncomplicated initial situation in which all sensors and measuring devices use the same interfaces and transmission protocols, both analog and digital sensors with an IO-Link connection can be found in the demonstrator. The particular challenge lay in synchronizing the transmission speeds in order to ensure smooth data transmission.

The measurement results for the current 'leakage scenario' can be read off the associated control system. They are also transferred to the Fraunhofer's own Industry 4.0 platform 'Virtual Fort Knox'.

© Mader

Usually, leaks in the compressed air system are still detected using ultrasonic technology as standard, as ultrasound makes the 'whistle' of the smallest leak audible. Although this is possible during ongoing production operations, it requires a corresponding amount of effort, as the detection has to be carried out directly on site. Nevertheless, Mader says that locating and subsequently eliminating such leaks is profitable - both economically and in terms of CO2 savings.

Fraunhofer IPA has set itself the goal of further optimizing the process. The demonstrator creates the basis for data-driven production research, for example by training self-learning algorithms. In future, they will not only be used to identify and localize leaks, but also to display the name and order number of the affected component via an app. This will help to minimize downtimes in particular.

The article was based on documents from Mader, Leinfelden-Echterdingen.