Please use this identifier to cite or link to this item:
Full metadata record
DC FieldValueLanguage
dc.contributor.advisorHoward, Carl-
dc.contributor.advisorGrainger, Steven-
dc.contributor.authorLarizza, Francesco-
dc.description.abstractBearings are widely used in rotating machinery to allow relative rotary movement, and the failure of bearings are a common reason for machine breakdowns. Machine breakdowns caused by bearings can be catastrophic, resulting in costly downtime or the loss of human life. An example of the severity of this problem is when a bearing in an axle assembly of a train seizes, it can cause the train to derail, destroying infrastructure and possible loss of life. The implementation of condition monitoring systems that use a vast array of methods to determine the condition of a bearing is used to reduce the risk of bearing failures. The focus of this research is on modelling defective bearings to predict the vibration response and estimating the size of a defect in a bearing. The latter is achieved by using the vibration response of an operational bearing and does not require the bearing assembly to be dismantled. Classical bearing condition monitoring trends the vibration amplitudes and the amplitudes of the vibration at bearing fault frequencies to determine if a bearing is damaged and needs replacing. In some industries defect size limits have been introduced that specify when a bearing must be condemned. Hence, it is advantageous to be able to determine the size of a defect in an operational bearing without removing the machine from service and disassembling it to inspect the bearing. Current defect size estimation methods include assumptions about the defect which are not necessarily representative of the real bearing faults, as well as not being able to determine if the size of a defect is greater than the separation angle of the rolling elements. This research aims to use the vibration or acoustic emission signature from a damaged bearing to determine the size and location of bearing defects. The research conducted and discussed in this thesis, used both experimental and analytical methods to examine the vibration response of a defective bearing with a spall defect on the inner and outer raceways, and dent defects caused by contaminants in the lubricant. An improved hybrid analytical and finite element model is presented that removes limitations of models developed by previous researchers. This is achieved by using a finite element method to determine the contact force and contact area of the rolling element to raceway contact interfaces, even when a rough surface is present. It was demonstrated that previous models that use Hertzian contact theory cannot model the change in the contact area as the rolling element approaches and exits the defect, as Hertzian contact theory assumes the surfaces are smooth and there are no changes in the curvature of the surface, which is not the case for a typical bearing fault. The improved model was validated using experimental data, and an actual defect profile that was measured. The improved model has been made publicly available. The research presents multiple methods to determine the size of a defect, and determine if the size of the defect is greater than the separation angle of the rolling elements. The first method involves band-pass filtering the time-series vibration response of a defective bearing to highlight the transition of a rolling element passing into and out of the defect. Next a spectrogram of the filtered response is used to determine when the rolling element begins to enter the defect and when it has finally exited the defect. The method was tested using ball bearings and cylindrical roller bearings under various constant applied loads and shaft speeds, in addition to having defects with various angles of sloped entry and exit edges. The method can accurately estimate the defect size under various applied loads and shaft speeds without requiring assumptions about the depth and shape of the defect. The second method was developed to remove a limitation of the previous defect size estimation methods, that cannot distinguish between a line spall defect and an extended spall defect that is caused by an apparent aliasing issue. The aliasing issue occurs because the time-series acceleration response is similar for both defects and is not because of poor data acquisition practices. The presented method uses the variation in the stiffness of the bearing assembly to determine if a line spall or an extended spall defect is present. It was analytically and experimentally proven that the characteristic frequency of the shaft-housing translation could be used to identify an extended spall defect, as the characteristic frequency is much lower for an extended spall defect when compared to the frequency of a bearing with a line spall defect. This difference in the characteristic frequency is caused by the fact that for an extended spall defect multiple rollers will occupy the defect area, so that a race is unsupported in that region, in turn making the bearing assembly less stiff. However, when the applied load is greater than a critical load, the characteristic frequency of the bearing assembly with an extended spall defect sharply increases to that of a bearing with a line spall defect, as the rolling elements in the defective area are now in contact with both raceways. Therefore, when using this method, the applied load on the bearing needs to be known to determine if the applied load is greater than the critical load. The final method that was developed to determine the size of a defect in an operational bearing is a modified technique that is used in structural health monitoring to locate defects in plates and shells. The proposed method uses the time difference a shear wave takes to reach multiple sensors positioned on the bearing housing and uses these time delays to determine where the wave originated. This method removes the limitation requiring the knowledge of the applied load and critical load, as required in the method that uses the varying stiffness of the bearing assembly, and can determine the location of the defect on a stationary component. This proposed method provides more information about the defect, such as the size and location of the defect, instead of only the estimated defect size available from other methods. The results of this study contribute to improving vibration condition monitoring of bearings by enabling the diagnosis and estimation of defect size in an operational bearing for line spall or extended spall defect, on either inner or outer raceways. All the experimental data and the modelling scripts are publicly available, which provides future researchers with an ability to confirm these results and data for future research.en
dc.subjectRolling element bearingen
dc.subjectDefect sizeen
dc.subjectVibration modelen
dc.titleVibration Signatures of Defective Bearings: Modelling and Defect Size Estimation Methodsen
dc.contributor.schoolSchool of Mechanical Engineeringen
dc.provenanceThis electronic version is made publicly available by the University of Adelaide in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. This thesis may incorporate third party material which has been used by the author pursuant to Fair Dealing exceptions. If you are the owner of any included third party copyright material you wish to be removed from this electronic version, please complete the take down form located at:
dc.description.dissertationThesis (Ph.D.) -- University of Adelaide, School of Mechanical Engineering, 2020en
Appears in Collections:Research Theses

Files in This Item:
File Description SizeFormat 
Larizza2020_PhD.pdf33.63 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.