Manual for Life Cost Based FMEA
Weibull analysis is extensively used to study mechanical, chemical, electrical, electronic, material, and human failures. The primary advantages of the Weibull analysis are its abilities to:
Provide moderately accurate failure analysis and failure forecasts with extremely small data samples, making solutions possible at the earliest indications of a problem.
Provide simple and useful graphical plots for individual failure modes that can be easily interpreted and understood, even when data inadequacies exist.
Represent a broad range of distribution shapes so that the distribution with the best fit can be selected.
Provide physics-of-failure clues based on the slope of the Weibull probability plot.
Although the use of the normal or lognormal distribution generally requires at least 20 failures or knowledge from prior experience, Weibull analysis works extremely well when there are as few as 2 or 3 failures, which is critical when the result of a failure involves safety or extreme costs. Parameters for the Weibull distribution for most components can be found in many references (http://www.barringer1.com/wdbase.htm).
Using distribution systems to model the MTBF in Excel will be discussed in the following section, Monte Carlo Simulation.
Mixture of Data
A system is most likely to be assembled with many components or subsystems. In order to predict an accurate overall MTBF, components operated under similar environmental conditions to the new system whenever possible. This means some subsystem may have empirical data and some may not. Using equation 12, we can predict the MTBF for any given system with data that from a combination of empirical data, manufacturer’s test data, and Weibull failure distribution. Modification factors in equation 6 are based upon testing and historical data. If the values are unknown, “1” will be used as the modification variable.
5. Monte Carlo Simulation
Life Cost-Based FMEA as described up till now uses point estimation for its analysis. The danger with using point estimation is the potential for misinterpretation of the average numbers. Strategy based on average conditions can be false since one does not know if the condition has reached the upper or lower thresholds. A sensitivity analysis on the estimates will provide better confidence in the result and make for a better understanding of which variables are the cost drivers.
FMEA MANUAL By S. Rhee and C.M. Spencer