Available Reliability Information
Abstract
Chris and Fred discuss how to deal with trying to find ‘reliability characteristics’ when you don’t have a lot of information. Sound familiar?
Key Points
Join Chris and Fred as they respond to a question about trying to find the reliability of something for a ‘high-reliability’ product in a systems engineering setting, focusing on electronic componentry through to PC housing. Wow! A lot to cover here. The question is really all about where we go to find information to help us.
Topics include:
- Literature. Especially conference proceedings of very ‘specific’ conferences. There are some gatherings of people that focus on very specific components and failure mechanisms. These can have a lot of information about how your component might fail … as long as you KNOW how your component fails. And then make sure the paper you find relates to your product.
- Vendor … carefully. Some vendors are actually across how their components fail. Some definitely aren’t. Some pretend or look like they are … but aren’t. But you won’t get what you don’t ask for.
- Similar products and components … knowing your assumptions. If you have been designing, building and producing lots of similar products in the past (previous models, generations and so on), then you might have a lot of information already gathered. But don’t just blindly copy information. Understand what makes your new product different to those from the past, and see what that means to the failure mechanism (yes … you always need to understand the failure mechanism).
- But ‘not’ parts count predictions. Jaguar and Toyota vehicles are functionally and technologically identical (or at least very similar). But Jaguar vehicles are much less reliable than Toyota vehicles (it’s not even close). So don’t just search for the ‘first’ number you can find!
- And if all this fails? … then you need to get your own information through things like Accelerated Life Testing (ALT), Finite Element Analysis (FEA) and so on. Physics will tell you if you need to do this. Not the finance team …
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
Sorry to pester you and rant, but GAAP requires statistically sufficient data to make nonparametric field reliability estimates for products and their service parts, using BoMs and gozinto theory, https://lucas-accendo-site-speed.sprod01.rmkr.net/gozinto-theory-parts-installed-base/#more-417514, if you’re willing to do some work and statistics.
https://lucas-accendo-site-speed.sprod01.rmkr.net/heres-data/#more-414861
This includes repairable systems, https://lucas-accendo-site-speed.sprod01.rmkr.net/kaplan-meier-estimator-for-renewal-processes/#more-531665. Not just MCF like Wayne Nelson preaches.
Ask your vendors for product and service parts’ ships and returns counts and in return offer the age-specific field reliability of the vendors’ parts in your products. I did that for HP but I can’t find the Accendo article that describes it.
… by GAAP I believe you mean ‘generally accepted accounting principles.’ The first problem with this … is that DDT, using leaches to treat ailments, smoking on airplanes, and a plethora of other substances and practices were once ‘generally accepted.’ And the second problem with this … is that we are not here to make accountants happy. All the things you mention above are completely valid ways of estimating reliability with statistics, but none of the data exists in the important early design phases or even pre-launch. Testing to get that data is expensive and time-consuming (to the extent you might miss your launch window for relevance). And even if you can test … it needs to replicate use conditions. Which we routinely stuff up.
There are plenty of organizations who routinely make reliability assessments without lots of data … and that is OK!