Acceleration by Grouping of tests

[bill_jones]

Since it is not possible to perform all of the environmental exposures on a product simultaneously, I am considering selecting a series of tests that each have approximately the same AF, which would only consume a portion of the products life for each group, and the repeatedly run the group of tests until failures occur.

Has anyone done this or something similar? How did it work out? Drawbacks? Tips?

Regards,
Bill

[David]

Hi Bill,

Interesting scenario. Do you actually know the AF? How many stresses are being applied during each test? If each test is focusing on a single stress then you could use the Cumulative Damage model within ALTA 7. For example, consider the following scenario:

Assume three stresses: current, voltage and force. The overall test is broken down into three phases:

Current = 100, Voltage = 0, Force = 0 for 300 hours. If the sample does not fail, the test continues to step 2.
Current = 0, Voltage = 100, Force = 0 for 200 hours. If does not fail again, the test continues to step 3.
Current = 0, Voltage = 0 and Force = 100 for 250 hours. If the unit does not fail again, a new cycle begins with step 1.ALTA does not allow a stress value equal to zero, so you could just use a very small number. Each step (1-3) would be defined with a stress profile. The total test time for all three steps is equal to 750 hours (300 + 200 +250). For the Current, the overall profile would be:

For the Current, the overall profile would be:

For time = 0-300, Current = 100; time = 300-750, Current = 0. This would be a cyclical profile.

Another option that I could recommend is to test one stress at a time while testing units to failure. In other words, one group of units would have Current as the applied stress, one with Voltage and the last group with Force. In this you assume that there are not any interactions among the stresses and that the failure modes generated based on the applied stress are independent. To obtain the overall reliability, you would then multiply together the reliabilities obtained from each stress. This is very similar to a competing failure modes analysis in life data analysis.

Additional information on ALTA 7 can be found at http://www.reliasoft.com/alta/index.htm and a free demo can be downloaded from our Download Store at http://www.reliasoft.info/download/product.php?productid=9&cat=6&page=1. Information on the Cumulative Damage model can be found at http://www.weibull.com/AccelTestWeb/time-varying_stress_models_chap_.htm.

I hope this helps.