Recently I got caught up in a reliability discussion between a user and a vendor.  The user worked for a large O&G company, and the vendor worked for a large company who manufactured process instrumentation.  In this particular exchange, the user was interested in the vendor’s safety shut off valve package.  While acting as a neutral party, I was concerned when the user asked the vendor about his package failure rate data, and the vendor confidently stated “181 years for the electronics and 60 years for the mechanicals”. 

Of course, as an accomplished safety engineer, I noted the old adage “Well, since we're only as strong as our weakest link, the answer must be 60 years, right?”   But to my surprise, the vendor came back with “really it doesn't matter, since there are no mechanical valves on the market that don't require maintenance more frequently than 60years” Huh?

I wanted to throw my ISA Book, “Safety Shutdown Systems; Design, Analysis and Justification” at him.  The co-author (and my good friend) ) Mr. Paul Gruhn, provides an enlightening section called “Metrics” where he describes all the ways we engineers manage to describe safety system performance (availability, reliability, probability of failure on demand, etc.) yet consistently tend to be misunderstood.   

Since I pride myself on clear communications; please allow me to explain one thing about failure rates.  Failure rates are simply a calculated guess on how long something is expected to work before it fails (yeah, I could have said “probabilistic determination” but I’m trying to get my point across).  Bottom line, it’s not a guarantee, just a level of confidence on how well something is expected to work.  So when your vendor tells you 60 years, just remember what he’s saying is that during a 60 year window, the device is likely to fail.  Unfortunately, he can’t tell you in which year that would be.  The key here is to “widen” the window.  Give yourself a better chance at success, ask the questions, and make sure you understand the answers…