Reliability ≠ Testing
Abstract
Mojan and Fred discuss the idea that a product’s reliability is not the same as doing reliability testing.
Key Points
Join Mojan and Fred as they discuss the importance of not confusing testing with a product’s actual reliability.
Topics include:
- A product’s reliability exists whether or not we perform any testing.
- Reliability testing can provide meaningful insights and information when done well.
- Before any testing, focus on designing a reliable product.
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.
/
RSS Feed
Show Notes
Hi Fred and Mojan,
I agree, design a reliable product first! And if possible use analysis to show you will meet the reliability targets. However, I do think testing is still necessary. I’ll try to explain where I am right now.
We are currently developing a complex system with some interesting chemistries (no we can’t just call up a chemist and have an answer in a day). We can develop models for the system and show that the degradation of the subsystems meets our targets, I’m with you on that front. However, in my opinion, there is no way I would still sleep well at night going to launch with our modelling alone. I still think a system level test with the time we have available is a good demonstration that our system is not only functioning as intended and we can identify some early life failures (infant mortality) with root cause activities, not only those things but we would still be better off having tested the system in reality than only relying on modeling. As I often hear, ~50% of failures come from interactions and modeling those interactions can be difficult at times. In our case, our customer has been dealing with competitor’s products with very poor reliability, so it is imperative that our product hits the marks here.
Would you agree with this? Am I right in thinking that testing is really required here to supplement and ground our modeling?
Hi Maxamillion,
thanks for listening and joining the conversation.
I’m not saying testing is never to be done, it’s should only be done deliberately when the outcome will add value and improve decision making. Testing to verify a model is prudent in most cases, testing to uncover issues that are not modelable makes sense. Testing for the sake of testing, does not.
cheers,
Fred