When a part breaks unexpectedly, it usually sets off a flurry of activity. Often, a team is formed and charged with finding the root cause of the failure. In my career, I have led or been a member of many failure analysis teams. Based on my experience, there are five mistakes that engineers often make when investigating failures. (I know because I've made them all myself.) Recognizing these mistakes can help you avoid them.
1. Looking at a part in isolation. As a materials engineer, people often send me a broken part in a cardboard box and ask, "Why did it break?" Of course, if the part had been in a cardboard box the whole time, it's unlikely that it would have broken.
Although there are many things that can be learned by inspecting a failed part, the part itself rarely tells the whole story. It's important to understand the mechanical system to which the part belongs and the part's role in that system. What loads does the part feel? Where do the loads come from? What might cause the loads to be higher or lower? It's also important to understand the environment in which the mechanical system operates. In what ways might the operating environment differ from what was anticipated in design?
Remember, no part ever fails by itself. It fails as a part of a given mechanical system under a given set of environmental conditions.
2. Focusing on conformance to specifications rather than root cause. When a part breaks, one of the first questions asked is, "Did it meet the print?" Of course, it's important to establish whether or not the part conformed with the design requirements.
But finding a non-conformance is not the same as finding the root cause of the failure. If a part is found to be defective, it's important to understand what role (if any) the defect played in the failure. In some cases, a defect may be a red herring, which leads you away from the root cause. In other cases, early failure of a defective part may reveal an underlying problem, which would also cause non-defective parts to fail over a longer period of time.
There's another reason not to focus excessively on the issue of conformance or non-conformance: It can result in a confrontational relationship with parts suppliers. This can undermine cooperation, which may be necessary for the success of the investigation. Failure analysis should always be about solving problems, not assigning blame. Of course, this is not to say that suppliers shouldn't correct non-conformances, or that there shouldn't be consequences for suppliers who consistently provide non-conforming parts. But this should be separate from the work of the failure analysis team.
As well as the fact that often a failure may have more than one root cause. Carpet bombing may be the best way to improve the system as a whole. Often you need to get the part to print. Make the design better so the tolerances can be larger. And improve the tool so the part doesn't vary as much. Anyone can design a part that has no tolerances and has to be made to print +/- 0.000001. But the truly good design engineer makes a part that can be made to +/- 3 mm. or bigger.
I liked the article as well. Currently I am find myself trying to get to the bottom line of a lot of failures. I find your article intersting because some of the things you suggest not to do are exactly what we are doing. Our focus tends to start by understanding how big is the problem. Not because we don't want to fix everything but more from the point that we don;'t have unlimited resources and we want to get the most bang for the buck. We tend to try and get data and group the failures into different root causes. And then do focus on if the part is to print. Quite often the failures are caused because the part is not to print. Once the part is to print and the variability is taken out the system then the root cause failure of the design can be attacked and improved. However, if the parts are not capable and can't be to print, it doesn't matter how good the design gets because you will still have problems.
And yes I have been in all of the above situations. My favorite is getting a part in a box and being asked "why it broke?" Only the part is fully functional....
While all these were sound advice I personally still keep an open mind for problems that would be fixed quickly by one of these sinful actions. Countless times I have attached 100 probes and just measured data... and wala ten minutes later I know the solution.
It did bite me once when the issue was not design or a problematic part but rather EMI. See probes can make the EMI issue go away...
Dave: No doubt this could be a case of finger pointing at its finest. I think the points you made are critical for engineering teams to sit back, take a deep breath and dive into the problem rather than attack it without a plan.
I am so glad after read this article. I am an Engineer and faced this problem many time and this article really helps me to solve my mistakes. Everyone should read this blog.
These principles also hold true for failure analysis in electronics. I worked for a semiconductor company for years as a product and test engineer and recognize most of these scenarios as having happened at one time or another. One of the most interesting places in a semiconductor plant is the F.A. lab which is usually where customer returns are evaluated. And of course when parts started failing on the production line, the first place everyone tries to blame is the test set - it never occurs to them that their process might have shifted...
@Beth: Thanks for your kind comments. You're right that none of this is rocket science; it's just rational thinking. However, when parts break, people understandably get upset. Emotions can run high, and there may be a tremendous amount of pressure. As Ann points out, under these conditions, even intelligent and highly educated individuals may start to behave irrationally. The most important thing is to stay calm and focused -- especially when others aren't.
@GlennA: Experts, in particular, are susceptible to the temptation to jump to conclusions. The more experience you have, the more likely it is that a given problem resembles one you have encountered before. But that doesn't necessarily mean it's the same problem! Sometimes experience can be just as blinding as ignorance.
@TJ McDermott: You're right that sometimes time constraints can force you into a "kitchen sink" response. However, in these cases, it may be a good idea to continue investigating even after the "kitchen sink" solution has been implemented in order to determine the real root cause. Who knows? Maybe you can make yourself look like a hero for a second time by coming up with a cost savings when you realize that 2/3 of the kitchen sink solution was unnecessary.
I agree with Beth. Dave, thanks for such a clear overview. The principles you discuss here seem simple and obvious in hindsight, yet somehow can be easily forgotten even by well educated and well trained pros. They parallel\ some of the basic electrical system troubleshooting principles I learned from one of my engineer buddies years ago, which I apply mostly to my multi-component stereo system.
Although I haven't been involved in formal failure analysis, I have often been called to troubleshoot problems. Often the most senior person involved 'declared' what the root cause was. After I finished my troubleshooting, I had often proven that the 'expert' was wrong. Seniority doesn't automatically mean that you know all of the intricacies.
On April 21, NASA launched a novel project, putting into orbit three satellites that employ an off-the-shelf commercial smartphone as the control system.
The legacy endpoint devices that control our critical infrastructure (utility systems, water treatment plants, military networks, industrial control systems, etc.) are some of the most vulnerable devices on the Internet.
In a switched-capacitor filter, capacitors and switches take the place of resistors and accurately reproduce the characteristics of continuous-time Bessel, Butterworth, and elliptical filters.
From Dell / Intel® New Paradigms in Design Work Scott Hamilton, vertical market strategist for Dell Precision workstations, 5/2/2013 5
Early in my career, I worked as a draftsman and remember the days of drawing on vellum with numbered pencils and Mylar with plastic lead. This was a fun experience in the sense that I ...
I've been using workstations for more than 10 years and love finding ways to get more performance from my system. With demanding professional applications that require more power each ...
A lasting memory from my first job as an engineer in an auto assembly plant is standing on hard concrete at six in the morning, vending-machine coffee clutched in hand, listening to ...
For industrial control applications, or even a simple assembly line, that machine can go almost 24/7 without a break. But what happens when the task is a little more complex? That’s where the “smart” machine would come in. The smart machine is one that has some simple (or complex in some cases) processing capability to be able to adapt to changing conditions. Such machines are suited for a host of applications, including automotive, aerospace, defense, medical, computers and electronics, telecommunications, consumer goods, and so on. This radio show will show what’s possible with smart machines, and what tradeoffs need to be made to implement such a solution.
To save this item to your list of favorite Design News content so you can find it later in your Profile page, click the "Save It" button next to the item.
If you found this interesting or useful, please use the links to the services below to share it with other readers. You will need a free account with each service to share an item via that service.
Tweet This
11 
