|
Activity Number:
|
147
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 7, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Physical and Engineering Sciences
|
| Abstract - #306151 |
|
Title:
|
Design of Sudden Death Tests for Estimation of a Weibull Percentile
|
|
Author(s):
|
John McCool*+
|
|
Companies:
|
The Pennsylvania State University
|
|
Address:
|
30 E. Swedesford Road, Malvern, PA, 19355,
|
|
Keywords:
|
sudden death tests ; maximum likelihood ; precision ; type II censoring ; Weibull distribution
|
|
Abstract:
|
Sudden Death Tests are a way of conducting life tests in which n specimens are divided into g groups each of size m (n=gm). Testing continues on each group until the first of the m failures occurs in each. The testing thus results in g failures among the n specimens. It is shown how to determine the group size m so that the p-th percentile of the Weibull distribution of life may be estimated with greater precision than in a conventional life test wherein n specimens are tested until the occurrence of the g-th failure. Comparisons of the expected duration of the life tests are given in both cases for several combinations of g, m and p.
|
