Five from 6 devices have failed

[raed]

Hi:
If i want to calculate the Weibull parameter (alpha and Beta) i have to perform the following calculation:

i sort the components anodically according to their failure points, then i calculate 'H'

H=(i-0,3)/(n+0,4), where: i is the rank n is the number of the components. Now, i had six devices under test but just five have reached there end-of-life.How i have to calculate H? (is n=5 or 6?).
thanks in advance

raed

[RS Support]

n=6 - if the last unit operated past the last failure time -- otherwise you need to use censored data methods..

See http://www.weibull.com/LifeDataWeb/right_censored_suspended_data.htm

You may also do this online at http://www.weibull.com/itools/index.htm

[DCraggs]

Assuming you are using a rank regression method and your data includes censored data, a good reference is "Reliability in Engineering Design" by Kapur and Lamberson, 1977, Chapter 11, Section 2.5 provides a detailed example on how to calculate ranks with suspended items.

[RAFA]

I want to calculate the reliability of our product, a simple mechanical assembly for the automotive industry.
No field data is available. It would be like predicting reliability for a new design.

In summary, best approach to predict reliability of mechanical assemblies, when no data is available.

[Adam]

Without data there are three approaches available:

1. Perform a reliability test.

2. Use historical data (i.e. old design), and then make some assumptions about the new design.

3. Use probabilistic design methodologies (i.e., predict probability of failure based on material properties, design dimensions, stresses, etc.).

4. Use the NSWC prediction standard (last resort). This approach should be followed by demonstration/validation testing.