WP1 R&D Mobility Fund | First Cold Spray Monitoring Trials

Imagine having an airplane part where some material has been removed, maybe by normal wear and tear, or maybe because some corrosion had to be blended out. Wouldn’t it be great if you could simply replace that material? Welding can do this, but welding also heats up the surrounding base material, destroying its fine-tuned microstructure and thereby actually making the situation worse. Instead, we can use a new technology, called cold spray.

Figure 1: The spray booth containing the cold spray system

Cold spray uses a supersonic stream of gas to deposit metal powder onto a surface. The force of the impact causes the powder particles to deform and bond with the surface they are hitting. This lets you build up a coating, replace material removed by for example wear or blending processes, or even create entirely new parts. Because the part you’re spraying on to remains relatively cool, the microstructure of the base material remains intact, offering a solution for parts that currently can’t be repaired.

While this is very promising, when repairing aircraft parts we of course need to guarantee the quality of the repair. How to do that in an efficient way? For cold spray, maybe the answer is…. to listen!

By recording the sound of the gas stream and analysing it with the help of machine learning, we can continuously monitor the cold spray process. This will let us detect any problems as they occur, or maybe even hear issues start to develop before they become a real problem. By combining our analysis of the sound of the process with an understanding of the material science, we may even be able to predict the quality of the resulting repair.

This is an ambitious goal, but this summer the Cold Spray Team within BrightSky took the first step, by conducting spray trials at Titomic Europe B.V. in Akkrum. During these trials test samples were coated, using different settings on the cold spray machine, while recording the sound. This produced a large dataset, which is now being analysed by Stratos Koufis, PhD candidate at TU Delft, to understand what information we can extract from this data, and what is the best way to do that. After that, the next step will be to link this process information to the quality of the repair.

Figure 2: Close-up of the microphone used to monitor the process

Will it be possible to detect nozzle clogs before they happen? Can we predict the hardness of the deposited material based only on the sound of the process? Keep following the BrightSky updates to find out how we progress!

Within BrightSky, the Cold Spray Team consists of EPCOR, HvA, KLM E&M, NLR, and TU Delft