Determining failure modes

[insua]

Hi , i want to ask about determining failure modes.
what i know about mixed weibull is; mixed weibull shows us different kinds of "failures".
for exp: parts have worn, broken problems. For worn problem, it is a failure mode and it can not be mixed weibull distribution or can it be???

but my datas for a "leaky problem" for a part;
the 2 parameter weibull line doesnot fit and it has a "S" shape, but when i do mixed weibull with three population, the result fits. What does that mean? Is leaky problem mixed distribution? and are there 3 different "causes" for leaky problem.
or 3 population mixed weibull means; there are 3 different "failures" that make my parts to leak???
thanks

[Pantelis]

A mixed Weibull model (see also http://www.weibull.com/LifeDataWeb/the_mixed_weibull_distribution.htm (http://www.weibull.com/LifeDataWeb/the_mixed_weibull_distribution.htm)) will allow you to create a single model that covers different failure rate behaviors (modes). Now and as mentioned a good graphical indication that a mixed Weibull model is present is when you get an “S” shape.

What does that mean: It means that the failure rate behavior (i.e. increasing, decreasing etc.) changes with the age of the product. The simplest example is when you look at the “idealized” bathtub curve where you have a decreasing failure rate behavior (early failures) followed by a constant and then increasing (wear out failures). Now for the case of your leak (and without really knowing too much about what it is and why it leaks), you are asking what does that mean. It depends. The following may be some of the reasons…


Different failure mechanisms (i.e. early, random and wear-out failures causing the leak)
Different failure modes (causes)
Differences in usage, stresses, application, testing and/or environment.
etc. …

[insua]

thanx a lot...