Life estimation can be determined by three different methods. The first and best process is the collection of actual operating life data that has sufficient information to determine the pattern of failures. This data, then can be analyzed using Weibull statistics to find the L-10 life point where 10% of the items have worn out. The next best process is to set-up a specific test program where the items are tested to failure and again apply Weibull statistics to find the end life point. The test program can be augmented by accelerating the failures by introducing higher level stresses. Acceleration factors will have to be developed to transform the test information to the normal conditions. The third way is to perform detailed analytical analysis on the sub-elements to determine the failure modes and mechanisms. These data can be used with life limiting experimental models to estimate the life times.
Mean Time Between Failure [MTBF] values have no relationship to useful life. MTBF is a measurement of the chance failures during the normal operating period with renewal of failures. Useful life is the time when wear-out failures occur.
There are lots of accelerated testing models, most are used when the failure modes are easily defined and the accelerating parameters are known. The best reference is Wayne Nelson’s book “Accelerated Testing”, ISBN0-471-52277-5. The RAC also has some information on accelerated testing, which can be seen at web address:
http://www.theRIAC.org/Publications/...eets/index.asp
05-22-2003, 02:39 AM
Life of product
Linear Mode
