Developing acoustic emission techniques for condition monitoring of rubbing contacts
File(s)
Author(s)
Gutierrez, Robert
Type
Thesis
Abstract
Sliding interfaces are critical to machine performance. Parts such as gears, bearings, piston rings and mechanical seals all have surfaces which are designed to rub against each other, often under high loads and running speeds. To ensure optimal performance and minimise economic losses, these areas must be closely monitored as this is where energy losses and failures commonly occur. This has led to rapid growth in the machine condition monitoring market in recent years. Acoustic emission (AE) is a passive monitoring technique which does not require contact with rubbing surfaces, nor transparent or conductive materials. However, research into condition monitoring with AE has been left behind other techniques. Studies have shown that AE is sensitive to tribological mechanisms, but the mechanisms producing AE at the contact are still not fully understood. This work looks to develop AE techniques for machine condition monitoring. Experimental work recording AE and friction from rubbing contacts under a range of conditions was done. Various data processing methods such as short time Fourier transforms (STFT) and short time histograms (STHG) were assessed, correlating AE data to friction coefficient and then applied to machine learning regression models to predict friction coefficient from AE data. Tests with Aluminium oxide coated specimens were examined, noting that AE can detect when coatings have been worn through. Scratch tests on these specimens were done, and by analysing wear track images and height profiles, it is shown that AE correlates with surface cracks and specimen plastic deformation. Finally, tests measuring AE from lubricated contacts were done. STFT processing combined with convolutional neural networks (CNN) classification models are explored to distinguish different oil additives and contamination levels present. Overall, AE has proven to be a powerful technique with potential to bring economic savings to industry by improving machine efficiency and preventing failures/unnecessary maintenance.
Version
Open Access
Date Issued
2024-10-17
Date Awarded
01/11/2025
License URL
Advisor
Reddyhoff, Thomas
Publisher Department
Department of Mechanical Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)