Manual for Life Cost Based FMEA
eliminate infant mortality rate. The failure rate during the early life period can be modeled by the Weibull Distribution (Ebeling, 1997).
Useful Life Phase: Beyond the infant mortality period, in the useful life period, the failure rate is assumed to be following an exponential distribution. The failure rate in this phase is at its lowest and is relatively constant. It begins after 10,000 hours (~1 year) of device operation. Failures during this
stage are due to unexpected or engineering can
random or normal wear and tear where failures are
sudden over stress keep the failure rate
at a negligible
caused by intelligent
3. Wearout Phase: Failures during this phase are due to component aging. Examples of what causes aging include fatigue, corrosion, creep, and other aging phenomena. Failures during this phase do not occur randomly.
Engineering work is done at the design stage to reduce or eliminate failures during the Useful Life and Wearout phases. FMEA allows us to identify failures in these stages for a particular design and to analyze how frequently these failures might occur, this will aid in increasing the reliability of the product. Three different approaches are discussed to predict the frequency of failures: Using empirical data, failure distribution estimation, and a mixture of both.
Step 1 Determine the new component or system that needs to be analyzed Step 2 Identify if such a component or system exists in the current system (SLC) If a similar or the same component exists in the current system follow section (4.1) Empirical data. If it is a brand new design then follow section (4.2) Distribution estimate. If some component of the total system is currently in use then follow (4.3) Mixture of Data. Step 3 Categorize the components or systems into different sizes, capacities, and features for the failure rate analysis.
4.1. Empirical Data
Many of the components designed for the ILC are very similar to designs in the SLC and PEP II SLAC used the Computer Aided Trouble Reporting (CATER) system from 1988 to 2003 and now is using Accelerator Remedy Trouble Entry and Maintenance Information System (ARTEMIS) to keep track of all component failures in the all the beamlines. Since we are interested in predicting the availability of the new system, only failures that actually brought down the accelerator will be considered as a failure
data point in the analysis.
A lot of information is
ARTEMIS database system and only the following an availability and frequency analysis: assumptions
captured points are
1. Date and Time of report The date enables us to ascertain beam lines were running when the failure occurred. Knowing the beamlines that were operational at that moment gives
FMEA MANUAL By S. Rhee and C.M. Spencer