We all know the simple paperclip. It’s just a piece of wire bent in strategic places, yet it's brilliant at keeping pages together. How often has that little clip saved my skin when a stack of papers clumsily slid off my desk? Not only is it super handy, it’s also super cool—at least for the Windows XP and Office 2003 generation, where "Clippy" was our trusted, if slightly eager, office assistant (“Siri” – we have come a long way!)

Nostalgia aside, that simple paperclip proved incredibly useful during a recent discussion about predictive failure rates and why certain prediction methods don't work for every situation.

Specifically, we were debating the difference between B10 data (which is based on highly accelerated life testing) and data from a Failure Modes, Effects, and Diagnostics Analysis (FMEDA), which is typically used in the process industries. There are plenty of blogs on this site explaining why solenoid failure rates predicted via the B10 method (cycle testing) don't translate well to an FMEDA methodology. FMEDA is far more appropriate for devices that are only cycled annually—or even less frequently—during a proof test.

The Crème Brûlée Challenge

At the end of a wonderful dinner, just as I was about to dive into my crème brûlée, I was challenged to explain why B10 data generally doesn't apply to process industry applications. Why was I droning on about different failure modes? Oh, and the kicker: I had to explain it so a non-engineer could understand.

Someone was clearly trying to keep me from my dessert.

Enter Clippy, our trusted paperclip.

Scenario 1: The B10 Method (Wear Out)

Consider a paperclip. Imagine picking it up, grabbing one end, and straightening it out—undoing the factory bend. Now bend it back, straighten it again, and keep going until it eventually snaps. We’ve all probably done this.

As you bend it back and forth, you’ll notice the metal change color slightly as it fatigues, leading up to that final brittle fracture. This is a perfect example of a highly accelerated life test. You are actively looking for the wear-out mechanism, which is exactly how B10 failure rate data is predicted.

Scenario 2: The FMEDA Method (Real-World Variables)

Now, let’s take that same paperclip. Instead of bending it back and forth continuously, we are going to bend it just once, place it on a desk, and wait 12 months before we bend it again.

At this point in the dinner, I was interrupted: "Dude, if I leave a paperclip on my desk, 12 months from now it will have completely disappeared."

"Exactly!" I replied.

In this second scenario, we encounter entirely different failure modes, such as cleaning or unintended disappearance. "Cleaning" might mean the paperclip gets tossed back into a drawer or thrown away. "Unintended disappearance" could happen when you accidentally drop it on the floor and the robot vacuum swallows it up.

The Verdict

What we have here are two different applications with completely different failure modes. B10 data simply doesn't apply to the second scenario because we will never actually wear the paperclip out. Even if we don’t lose it, we might just end up replacing it after 10 years at the end of its useful life. In this case, mechanical wear-out isn't a failure mode we should even consider when determining the failure rate.

As I finally started enjoying my crème brûlée, my challenger began to connect the dots on his own.

"Sometimes I clip papers together, put the folder away, and only look at it every few years," he mused. "I never really bend the paperclip. In fact, sometimes when I look at older files, there's actually rust on the paperclip. B10 testing would never account for rust."

Taking my last bite of dessert, I couldn't help but laugh. That versatile paperclip—Clippy—strikes again!