Comparison wizard

[JohnK]

I have a question on the Comparison Wizard in Weibull++ 7. When I compare two sets of data I get the probability that one set will out last another set. I would figure that if I switch the sets and got the new probability and added to the first probability that it should add up to 1. However, I don’t get this. Why?

John

[Arai.M]

They should in most cases (Manual bounds might skew results for example). However both solutions are calculated numerically and some convergence/round off error might be present.
Arai

[Tarik El-Azzouzi]

You can attach your file here or contact customer support if you think that the difference is significant.

[Steve]

I have two data sets I am comparing with the Comparison Wizard. I find that if I override the upper limit to a shorter time period (instead of infinity) I can get one data set "lasting longer" with a probability of 100% depending on how I select them. In other words, if I compare A to B then A will last longer. If I compare B to A, then B will last longer. How can this be possible?

[Arai.M]

Steve,
Could you either e-mail your file to support or attach it?
Regards,
Arai

[Steve]

I'll try to send it along. This isn't first time I have run into this issue.

[Steve]

Ok, back on this again. No need to send data because this example illustrates the point:

http://www.weibull.com/hotwire/issue125/relbasics125.htm

Comparing A to B, W++ returns:

"Folio1/Design B will last longer
with probability of 100%"

Comparing B to A, W++ returns:

"Folio1/Design A will last longer
with probability of 95%"

This is with upper limit on time = 1000 and quadratures = 1000 just as in the example.

I will continue to read through the example, but certainly you can see how this might cause some consternation in a roomful of engineers.

[David]

Hi Steve,

I can certainly understand the confusion regarding the different results, but hopefully the article helps to clear up the confusion about why the results are different and what the results in the Tests of Comparison actually represent.

Thanks!

[Steve]

Let me see if I understand the example.

For a warranty period = 5,000 hours, in simple terms:

1) A outlasts B 4.2155% of the time.
2) B outlasts A 63.3380% of the time.
3) Both A and B last longer than 5,000 hours 32.4465% of the time.

Is that correct?

[David]

Yes, you are correct.