Talking about MTBFs and PDFs with the Minnesota Reliability Consortium
The Minnesota Reliability Consortium is a division of the IEEE Reliability Society, and is also supported by the Minnesota Section of ASQ. They host a monthly presentation from reliability professionals on a wide array of topics.
Acuitas' own Chris Jackson delivered his MTBFs and PDFs presentation. The MTBF is a dangerous metric in terms of reliability, and you really need to focus on PDFs. Why? Because there are virtually no reliability applications where the MTBF is important. Whether we are a commercial entity that cares about warranty returns, a healthcare service provider that needs medical devices to work or a military that needs its weapon systems to work … we really only care about reliability at certain times or mission durations.
And that is why we need to know about PDFs. Chris started with everyone's 'favourite' - the bell curve. The bell curve describes random variables that cluster about a certain mean value. When we are talking about times to failure, this typically means that the system in question tends to fail 'away from time = 0.' In other words, the system is wearing out. So if you see a bell curve in reliability engineering, we typically deal with wear-out.
And when we use the bell curve to describe time to failure, 50 % of your systems will have failed by the MTBF. This is because the bell curve is symmetrical.
The MTBFs and PDFs presentation included an animation that showed that as the bell curve 'morphed' to PDFs that moved toward zero, the percentage of systems that will have failed by the MTBF. What this means is as the failure rate moves from exclusively wear-out to wear-in, more and more systems fail early.
If the failure rate is zero, then 63 % of your systems will have failed by the MTBF. And it gets worse if your system fails primarily due to infant mortality or wear-in. In fact, the more your systems fail due to infant mortality, the percentage of systems that will have failed by the MTBF move closer to 100 %. This doesn't sound intuitive ... but its true.
As a rule, most people don't truly understand this. For example, when engineers set servicing or preventive maintenance (PM) intervals, they tend to think about the MTBF. So if a particular gasket has an MTBF of 9 months … the PM interval is set to 9 months. But we know at least 50 % of them wil have failed by then! So very little failure truly gets prevented.
If you would like to learn more … please feel free to contact us!