Welding, crushing, breaking up concrete – an Empa team monitors noisy processes using artificial intelligence. This way, the sounds of production faults and impending breakdowns can literally be filtered out of the noise.
Science and industry has so far lacked a suitable monitoring method for identifying whether an electric discharge has actually hit the concrete or rock during electric pulse fragmentation.
Empa
Swiss-based Selfrag, from Kerzers near Bern, manufactures high-voltage generators that can break up concrete using artificial lightning. In contrast to a sledgehammer, which creates sharp fragments of concrete with split pebbles, this electric discharge can break concrete down into its constituents of gravel, sand and cement, enabling it to be fully recycled. However, up until now there has been no suitable method of identifying whether the discharge has actually struck the lump of concrete or rock. Kilian Wasmer and his team from Empa's Laboratory of Advanced Materials Processing in Thun, Switzerland, were tasked with developing such a method.
To achieve this, they bombarded small acrylic glass specimens with high-voltage blasts of artificial lightning. The acoustic signature of every blast was recorded and each specimen examined under the microscope for cracks and damage to the surface. Based on this data, the scientists finally managed to distinguish successful electric discharges from failed ones and also to detect surface hits. The researchers' success in analysing electric discharges in real time also inspired them to analyse other extremely noisy processes, such as the screeching or rattling of machines.
Correctly interpreting the screeching of machines
When not properly oiled, roller bearings and other moving metal parts may become jammed. This problem causes considerable damage worldwide. So far there have been no effective early warning systems: temperature sensors integrated in vulnerable parts only detect a temperature increase once the jam has already occurred and the parts have been ruined. However, it's not always necessary to completely overhaul a machine just because it makes a screeching sound. Dismantling and servicing production machines more often than required causes unnecessary costs. Waiting too long, however, may cause a moving part to become jammed, destroying other parts of the machine. It is therefore imperative to be able to pick out the critical screeching sound from the jumble of noises, so that the machine can be stopped before it is damaged.
Wasmer's team allowed a hardened steel roller to rub against a cast-iron base on a tribometer, an instrument for measuring friction. The team recorded the sounds and stopped the experiment at different stages in order to examine the damage under the microscope. The researchers managed to pick out the important clues from the chaos of noise generated by the roller on the cast iron. The team can now detect parts jamming with 80 percent certainty. The critical phase before the jam can be detected with 65 percent certainty, which would allow many industrial machines to be stopped early enough to protect them from serious damage.