No MTBF and Reliability and Maintainability Symposium
RAMS is an annual multi track technical conference, rams.org. This year in Orlando and about 400 or so attended. I was rather busy, having two tutorials, and three papers. See my site for a copy of my slides and papers. It was fun, exhilarating, and very enjoyable.
Last year at the Reno airport as I was waiting for the shuttle to the RAMS conference hotel, someone came up to me and asked my opinion on MTBF. He recalled jokingly that he had picked up a few NoMTBF buttons the previous year. We chatted about the issues with MTBF as kindred spirits. The rest of the conference I mentioned NoMTBF and distributed buttons, and had a few conversations.
This year I received a few emails prior to the conference and they mentioned the NoMTBF efforts. In particular I received permission to post the Grundfos internal policy to not use MTBF and quickly posted a note on that topic.
At the conference, I would estimate that 8 out of 10 conversations concerned MTBF. One asked, why I was so adamant? another asked for advice on convincing coworkers, another asked for guidance when the entire industry ‘required’ MTBF. And, many of the conversations started with a story of how bad it was where they worked, or how poorly those in their industry mis understood MTBF.
A long way to go remains
On the other hand, a few papers still used MTBF as the means to document reliability. Sadly these did not even come with a statement of duration. We as reliability professionals should do better.
First, don’t use MTBF, and if required to make a statement using MTBF, include a duration over which the MTBF value is valid. That should confuse the ignorant into asking a question about duration and probability of failure.
Second, when you notice someone else using MTBF ask them what they mean by that statement. Ask them to clearly explain why the assumption of constant failure rate is justified. Ask they if the data hidden by the averaging may contain useful information. Ask them to not use MTBF as it is too confusing, misunderstood and often inaccurate.
Third, If you have a story about successfully not using MTBF in the face of industry and peer pressure – lets celebrate you work. Send a short note and permission to post here or add a comment. Help to bring hope to those still struggling against such a poor metric.
Fourth, if you have a great story about how bad the use of MTBF can be likewise, let’s share that here too. I know of millions of dollars wasted due to reliance on MTBF. A simple change to R(t) or reliability being the probability of success as a function of time revealed the waste and changed the organizations behavior.