Ranking of Fatigue Data Based upon Monte Carlo Simulated Confidence Number Figures

Since fatigue is probabilistic, trends observed in large populations of data are necessary to select materials, compare engineering designs, or establish preventative maintenance schedules. The generation of large experimental fatigue populations, however, is prohibitively time consuming and costly. As a solution a Weibull-based Monte Carlo simulation of fatigue life was developed based upon a failed "bin" model, and five billion fatigue lives were simulated. These fatigue lives were used to generate L10 lives. A model of confidence number was developed dependent upon statistically large samples of simulated L10 fatigue lives, and independent of a limited number of published curves. Using these simulated values, Confidence number figures were generated that deviated from 0.0% - 7.4% of previously published figures and were independent of confidence bands. Results differed as little as 1% from those determined graphically for experimental bearing data sets while graphical interpolation was eliminated.

Name thesis:degree_name

Master of Science in Applied Engineering (M.S.A.E.)

Level thesis:degree_level

Thesis (open access)

Discipline thesis:degree_discipline

Department of Mechanical Engineering

Year dc:date.available

2011

Author dc:creator

• McBride, Jacob

Contributors dc:contributor

• David Williams
• Aniruddha Mitra

• ETD
• Fatigue
• Preventive maintenance
• Weibull analysis
• Monte Carlo analysis
• Probability analysis
• Rolling element bearings
• Population comparison
• Population ranking

Repository record dc:identifier

https://digitalcommons.georgiasouthern.edu/etd/772

OAI identifier oai:identifier

oai:digitalcommons.georgiasouthern.edu:etd-1772