[ | Next | Previous | Up ]

Re: Reliability in the Automotives

From: Ken Poi
Date: 12/20/2004
Time: 5:42:38 PM

Comments

Hello Stan,

A Design validation test plan requires samples to run specific tests from a test matrix. A similar phase in production called Ongoing reliability testing, will require the run of more production runs in reliability testing on a subset or even similar tests to Design Validation.

Question What strategy should I use in defining the sample sizes given the reliability objectives being set at 0.97 at Confidence level of 50% ( could this mean small samples are allowed ) to a MTBF spec of 150000 hours.

My task is to select a (fixed) sample size, justify it statistically, determine the frequency of taking samples ( a figure of 3mths was tossed in for the discussion ). All this isnt acceptance sampling but reliability testing with fixed sampling for fixed intervals. The objective is to assess the Production's reliability ongoing basis throughout the life cycle of a product in production.

In the past disk drive industry, I have worked and been exposed to SPRT test plans for product and process reliability assessment. but now in the automotive industry, testing is done on a matrix of small samples testing, like 2-6 units per test and units continue on to do other forms of testing if they pass one test. So some units flow through 4-5 tests (non destructive tests) before they finish the project ( lasting 3-4 months) . My questions are, How do I quantify the complete (cumulated) sample size, frequency.

Assuming I use alpha = 0.05 Beta = 0.05 can I calculate n over periods of months.

I am more at ease of using PRST sampling plan to demonstrate 150000 hours with weekly samples calculated to be about 30-50 per week using a discrimination ratio alpha/beta = 2 and confidence level of .90.

Not sure if I can start a new discussion thread here without asking you to do the job I am tasked to do. Just fishing for more technical discussion on the subject before I cast in concrete the plans I have posted already.

Regards Ken Poi Reliability Test Engineering


Last changed: November 20, 2007