Field and Operational Time

Abstract

Chris and Fred discuss a listener’s question about dealing with different ‘duration’ metrics for their field data … so how do we find things like ‘operating time’ and other age-related metrics that aren’t calendar time that simply aren’t there?

Key Points

Join Chris and Fred as they discuss how to work with a product’s whose age and usage are only available in terms of calendar time, but we are interested in things like ‘operational’ or ‘running’ time instead. Help!

Topics include:

Before you launch into analysis … what decision are we trying to inform? Focus on what you’re trying to achieve or inform. What decision is this for? Working out this decision lets you know how accurate the outcomes of your analysis are. If you only need an answer that is a ‘rough order of magnitude’ … then maybe you don’t need to waste too much time on this problem.
What does this mean? Here’s an example. An engineer tested three DC converters from three different suppliers for three times the anticipated service life in terms of ‘running time’ for their complicated machine. Two of them didn’t fail at all. One failed, but right at the end of the test. Long story short … none of these three DC converters would cause the machine to fail if they were selected. So which one is the right one? The cheapest/sexiest/thinnest/lightest/one with the lowest supply chain risk. They are all ‘reliable enough’ for the machine … so there is no need to even use reliability to differentiate between each.
But if we need a more accurate number? Then you need to do some digging. And depending on how deep you want to go … some more complex data analysis. If you have access to data like shelf life before sale, observed usage (odometer) readings from products being used and so on, you can come up with a distribution that describes the likely values for operational time based on calendar time. So if something fails after 300 days, you might be able to create something like a ‘bell curve’ that tells you the best guess at running time is 195 hours, with us being 90 % certain it is between 175 and 215 hours. Or … you might need the marketing team to launch some surveys and do some research. And if you know what the decision you are making is, you can work out how much money could be made or saved by making the right decision. And that might help the justification for things like surveys.
Still not happy with this? What happens if you make the worst ‘assumptions’ regarding your analysis, so the analysis gives you ‘worst case’ and ‘best case’ numbers. If either of these numbers give you enough information to make a decision … then you are still good to go!
Just focus on the decision …

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.