Date of Award
Master of Science (MS)
Ball bearings with solid lubrication lack the damping mechanisms of oil and produce well-defined vibration characteristics based upon given geometry and speed of operation. This work takes advantage of the high signal to noise ratio in x-ray bearings and develops an algorithm to statistically track bearing performance based upon fundamental bearing theory, Monte Carlo simulation, Order Analysis, and Weibull statistics. The technique gathers vibration data solely related to the theoretical operation of a bearing, negates the background noise, and provides descriptive vibration amplitude statistics of individual bearing components for evaluation. The practical implications of the thesis described herein allow the bearing engineer to optimize designs for life and noise by essentially tracking bearing component condition during operation. The output of the research is a tool/methodology to study and describe the vibration pattern of a bearing in operation. The concepts are tested and verified through the creation of a vibration transfer function between a sub-assembly bearing test rig and a bearing at full assembly level. Developed with Weibull statistics, the function correctly describes populations of vibration amplitude approximately 50% of the time within a 72% confidence interval. This was possible because the statistical methodology created found physically meaningful vibration patterns, not random vibration patterns.