Balancing Reliability Testing and Modeling
Andre and Fred discussing the interplay of modeling and testing to enhance the value you get with your testing.
Join Andre and Fred as they discuss the ASTR panel discussion on the topic of reliability modeling and testing.
- Is it possible to model so well that you can avoid testing?
- The lack of useful information in standards-based testing
- Doing testing to compare to expectations so you can learn something
Enjoy an episode of Speaking of Reliability. Where you can join friends as they discuss reliability topics. Join us as we discuss topics ranging from design for reliability techniques, to field data analysis approaches.
SOR 157 Balancing Reliability Testing and ModelingAndre Kleyner
Arvind Vasan says
I agree with most of the discussion with Andre (he is friends with me), except for one item where you mentioned that Prof. Pecht talks about modeling and eventually not doing testing. I am sorry, but this is totally wrong. I graduated from CALCE and I have spent a lot of time with Prof. Pecht.
His idea and so is mine is not to use modeling as a replacement to testing. PoF-modeling helps in correlating field to Accelerated test conditions.
As you were mentioning in this talk, a lot of test are simply designed for certain number of hours and if devices meet this test duration the devices are labelled as passed and qualified.
PoF model uses the field loading conditions, material properties, and design geometries to estimate the damage likely to accumulate in the field conditions. Then for the test conditions, we use the same model (but in a backward manner) and identify the test ‘minimum duration’ to accumulate the same amount of damage. If we adopt this approach, we will know the minimum time duration to test in order to get confidence in meeting a certain reliability in the field.
PS: Its really hard to type every word. If you would be willing, then we can have a phone conversation.
Fred Schenkelberg says
Thanks for the comment and insights. While not having a close contact with Michael, I’ve heard him and team, as well as the folks at DFR Solutions suggest the PoF models can be used in lieu of testing, based on the testing that has gone into creating the models.
I agree knowing the full range of conditions is often difficult limiting the value of PoF models. Not impossible, yet the less we understanding how and where a product will be used, the more assumptions we will have to make. This is true for testing or modeling or both.
Given limited time and resources for reliability testing, would you use a well-vetted PoF model in lieu of testing? If not why?