PDA

View Full Version : Mixed Weibull, multiple failure modes in ALTA 7 Pro and Weibull++7


JJPEngr
June 3rd, 2008, 09:59 AM
We have recently encountered multiple failure modes in two analysis cases and available articles on weibull.com or online help do not address what we need to do.

1. in ALTA 7 Pro, a set of data has two failure modes. How do we use Weibull++7 to do a mixed Weibull and then go back to ALTA Pro to plot separate life stress relationships for each sub population?
The particular data set has 13 samples, three stresses (voltage, load current, and duty cycle) with 2 failures of one mode and 5 failures of a different mode.

2. In a set of field data with a large population (several thousand samples and mostly suspensions) over a year and a half, Weibull++7 with a mixed Weibull shows 2 clear subpopulations with betas of about 0.7 and 2.0. We know of at least three predominant failure modes. Is there a way to specify subpopulations in Weibull++ 7 rather than just let the tool select them? With such a large population, is there an easy way to separate out subpopulations with each of the predominant failure modes for separate analysis?

Thank you

David
June 3rd, 2008, 06:27 PM
Hi JJPEngr,

For question #1, if the values of Beta for the two failure modes are not statistically different then you can combine all of the data and analyze within ALTA. If the Beta values are statistically different then you can analyze each failure mode separately. For example, when analyzing FM1, any failures due to FM2 would be treated as a suspension. When analyzing FM2, any failures due to FM1 would be treated as a suspension. This is basically the method of competing failure modes (http://www.weibull.com/LifeDataWeb/competing_failure_modes.htm). This assumes that the modes are independent of each other.

Question #2 has been answered at http://www.reliabilitydiscussion.com/showthread.php?p=3895#post3895.

I hope this helps.