PDA

View Full Version : No failures occured in a test


probsmth
September 5th, 2008, 06:17 AM
I have to evaluate a test which did not show any failures during testing time, the individual testing times are different to each other. Thus I assessed them as "suspensions".
W++ allows calculating a 1-parameter Weibull distribution with an assumed form parameter beta at a given confidence level. I am familiar with that.
Now to my question:
I would like to calculate a Lognormal distribution with these suspended parts with an assumed log standard deviation at a CL 50%. Unfortunately W++ does not allow this and proposes using a 1-parameter Weibull. Can anyone explain me, why? :confused:

My interim solution is calculating the suspended parts with a 1-parameter Weibull (beta from experience), calculate B50 and take ln(B50) as my-log for the Lognromal distribution.

David
September 5th, 2008, 12:16 PM
Hi probsmth,

I apologize for any inconvenience given that the 1-parameter lognormal distribution is not available. I have added the suggestion to our internal system and we will look at adding this feature into a future version of Weibull++.

Your comments are greatly appreciated.

probsmth
September 7th, 2008, 10:17 PM
Hi David,

Thank you for your immediate feedback, it sounds good to me.

Best regards, Mathias