Intermittent Failure During a Test
Abstract
Chris and Fred discuss how you analyze the failure data from a test … where failure is sometimes intermittent?
Key Points
Join Chris and Fred as they discuss intermittent failures during the test. What do you do? For example, let’s say something is being tested to see if it works every ‘cycle’ of operation. The idea is that it needs to last a certain number of cycles to work. What happens if the item passes the test … but there were a couple of cycles during the test where it didn’t work – but it worked for the next cycle?
Topics include:
- If you need to conduct Root Cause Analysis (RCA)? … then intermittent failures mean you there is often no physical trace of failure. So what do you need to do? Ensure that you use deductive RCA brainstorming to identify probable root causes, prioritize them for Corrective Action (CA), and see what happens.
- What is a ‘failure’? Will the customer care if the item works on the cycle after it fails?
- … however – beware of behaviors you don’t understand. This could be early warning of much bigger problems. Not sweeping these behaviors under the carpet has often led to discovering issues that might develop further when being used.
- So what do you do in terms of statistics? You (perhaps unfortunately) need to focus on what you defined failure to be, and what the pass criteria were at the start of the test. If there is no room for intermittent failures in any definitions of ‘failure’ up front … then statistically, you need to ignore them. However! Statistics are based on assumptions, even if you think you don’t make them. An assumption you make during testing is that the thing you are testing is representative of the thing that will be used. And you can’t understand where those intermittent failures come from … you can’t make this assumption. Until you do … statistics mean nothing.
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.
Leave a Reply