Abstract
The probabilistic nature of fatigue life requires that the scatter in the data must be considered when establishing an acceptable life and the inherent risk associated with that life. The impact of the test population size (number of failures) upon the scatter associated with fatigue data sets must also be understood. The authors have successfully used Weibull-based Monte Carlo simulations were successfully used by the authors to predict the fatigue lives of small populations of rotating shafts, gears, rolling-element bearings, and transmission systems. These methods have been based upon randomly drawing failed parts from a virtual bin of parts. To varying degrees, these methods have been validated with limited experimental results. The work reported here generated a large number of experimental rotating aluminum shaft fatigue failures (357), from which sub-populations were can be randomly drawn, replacing the virtual bin in the fatigue life method. This is referred to as a hybrid Monte Carlo-Experimental approach Similar trends in variability in L10 life with population size were observed with fatigue lives generated using the virtual bin method or pulled randomly from the experimental population. The Weibull-Johnson Virtual Bin Monte Carlo Fatigue Life Method for Rotating Shafts was proven to be a reasonable alternative to determining fatigue life when insufficient failure data are available (in essence, experimental vs. virtual-bin data corroborates the methodology).
Original language | English |
---|---|
State | Published - 2012 |
Event | 14th International Symposium on Transport Phenomena and Dynamics of Rotating Machinery, ISROMAC 2012 - Honolulu, HI, United States Duration: Feb 27 2012 → Mar 2 2012 |
Conference
Conference | 14th International Symposium on Transport Phenomena and Dynamics of Rotating Machinery, ISROMAC 2012 |
---|---|
Country/Territory | United States |
City | Honolulu, HI |
Period | 02/27/12 → 03/2/12 |