Many Ways to Estimate Reliability
Abstract
Chris and Fred discuss how we can go about estimating reliability … especially when you don’t have a lot of data, experience or other things that textbooks say you need to have to estimate reliability?
Key Points
Join Chris and Fred as they discuss how you go about estimating reliability, especially if your product or system is not mature or well-understood. How do you measure reliability early in the design process (for example)?
Topics include:
- Start with the decision you are trying to inform. For example, if you are trying to work out which gasket or material is more reliable than another to inform your supplier choice, then you don’t need to have an incredibly accurate estimate. Are you instead trying to work out what specifications need to go into a document for a supplier’s component performance, then you simply might need a ‘best guess’ of all the other component reliabilities to get that right. And the less confidence you have in the ‘best guess,’ the more margin you might need in that specification. Is this going to be a problem? If not, then don’t bother doing more modeling or analysis.
- … so it is about the ‘confidence’ you need. Not every reliability measurement needs to be accurate to six significant figures. Focus on the decision first, and then actually talk about the confidence you need so that people don’t panic over getting a ‘perfect’ answer.
- Estimating reliability is primarily there to provide guidance. Too many organizations are all about having a number that can be put in a document, brochure or meeting minutes. Let’s say that we come up with a ‘best guess’ for a component at the start of the design process. As the design becomes more mature, do they still feel good about this number? If not … talk about it! Can this be accounted for in other components? Or will this break the bank? Just start thinking (and talking)!
- And don’t use parts count predictions … we have lots of webinars and podcasts on why not!
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
I admit I skimmed your audio tape. Did I miss your discussion of estimating field reliability, without unwarranted assumptions??? Credible Reliability Prediction???
I agree that thinking about what people will do with reliability information is worthwhile. Armchair exercises are not enough: FMEA, FMECA, RCM, RPNs, MTBF prediction, “Reliability” growth (really MTBF growth), ???
->How can people get away with doing “reliability” work without knowing field age-specific field reliability and failure rate function for all products and service parts, including past products and parts?<–
Professional organizations like ASQ RD and IEEE Rel. Soc. promulgate and prolong traditional practices. Schools and training companies still teach the statistics I learned 50 years ago. (Wait for my next article!)_The military and Defense Logistics Agency still practice actuarial methods ONLY for gas turbine engines with lifetime data.
Sorry to rant.
When I worked for Agilent, I found a data base had lifetime data for service parts. I needed it and used it (to make nonparametric estimate of field reliability and failure rate functions) for interferometers. Infant mortality was evident and probably due to process problems. The process engineers agreed. I offered to do the same for all Agilent service parts. Not interested.
https://sites.google.com/site/fieldreliability/would-you-believe-you-dont-need-life-data/artificial-reliability-is-there-reality-in-reliability
Hi Larry … there is a lot to unpack here! I think you did miss that a little bit. The big issue here is that many people simply don’t want to take the time to understand HOW their product or system will fail. The big organizations you refer to (especially the military) cannot tell you how their products will likely fail. This is something the process beats out of them (admitting things can fail is seen as … failure). But the organizations who do things well know what will ultimately cause the failure of their product (and be comfortable talking about it). You need to know this first … and that means you can’t simply find the answer in a Google search of a book of numbers.