Reliability Prediction Standards
Abstract
Dianna and Fred discussing the history and application of published parts count prediction models and standards in reliability analysis.
Key Points
Join Dianna and Fred as they discuss the history of the parts count prediction models that are published for sale or that used to be publicly available but have since been withdrawn.
Topics include:
- A story of how these standards were created
- Why it and similar standards are out of date
- Why some people still ask for it
- What to do instead
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.
- Social:
- Link:
- Embed:
Show Notes
Published parts count prediction models: please don’t use these. These models are not useful, and they are misused.
Some of the reasons these models are out of date:
- They’re simply not maintained.
- Data collection is a challenge.
- Companies do not share their information.
- Companies are competing for best products. Standards lump all products together.
- Companies are concerned about litigation and do not want to publish the reliability information of their products.
- Data listed in the standard are not a reflection of field reliability.
People may still request the parts count reliability because of holdovers on procurement forms. If the calculation is not going to be used to determine any kind of reliability information, then we can provide it. Otherwise, don’t use part count prediction models just because it’s simple to do.
Field reliability depends on your parts, your designs, and your customers. Users need to maintain and update these reliability models, otherwise they become irrelevant.
It’s difficult to find really good TTF (time to failure) data at the component level. From vendor-to-vendor we see widely different claims. And companies are reluctant to share information because of trade secrets.
Some companies perform reliability testing on their components. They may report it as tests of the weakest component in their system.
If it’s important for your designs:
- Create a block diagram. Try to get the distributions NOT failure rates, FIT Rate, or MTBF. Today’s software is powerful enough to be able to use block diagrams with reliability distributions.
- Ask Vendors: “What’s the TTF distribution and how do you know, and what’s the failure mechanism?”
- Or partner with vendors to test critical components. There are many mutually beneficial ways to work with vendors so you can get the reliability data you need.
Leave a Reply