A product failure analysis is an activity manufacturers hope they’ll NEVER need to do.

If something has gone wrong enough to warrant doing one, you’re probably experiencing product quality or safety issues that are so severe that customers are returning them in numbers, or even, at the extreme end of the scale, getting hurt!

If you are hit with quality, reliability, or safety problems with your product during development or, more likely, in the field, it may be necessary to help you pinpoint the issue, decide what kind of corrective actions are required, and contain the problem. However, it must be said that it’s not a popular activity amongst smaller businesses, perhaps to their detriment. 

Luckily, there is a tried-and-tested product failure analysis methodology you can follow…

Listen to the post here:

🎧 How to do a product failure analysis?

Why many factories have an ineffective ‘firefighting’ or ‘scattershot’ approach to problem-solving that doesn’t work well.

Factories often scramble to fix a problem but don’t end up checking results to confirm if it worked or analyzing and understanding why the problems occurred. Often the boss or engineers jump to a conclusion with no analysis based on a hunch from their experience which may actually be the wrong fix and this gets carried out, solving nothing. If a component supplier made a mistake, for instance, your manufacturer may just claim they’re going to get tough on them, but there’s no system in place to do so, therefore the customer has no idea if this issue will reoccur or not. Furthermore, no record may be kept of fixes meaning that in the future staff have nothing to refer to if they encounter the same or similar problems (assuming that a fix was right in the first place). (01:18)

What kinds of problems warrant performing a failure analysis?

Failure analysis may be required in these 2 scenarios:

During new product development

prototypes are made and they might not function as planned and fail in an unexpected way. In simple products (like playing cards, for example), the cause is often easier to find and experienced teams can jump straight to a fix without doing failure analysis. For complex products or products where failures could be dangerous, such as medical devices, engineers doing a structured failure analysis is preferable at this point. The analysis needs to be documented, used when developing the next build, and implemented in the next round of prototyping which will then be checked to validate that the problem is now solved. (06:27)

After products have been shipped

Most of the time failure analysis is done after products have been in the field for months and product reliability issues occur leading to user complaints. Reliability issues usually appear later and often stem back to product design deficiencies, but swiftly making a design change to fix the problem without re-testing, verifying, and validating it can cause even more issues. It’s better to do a failure analysis first to understand the causes of the problem by looking at all possible information and deciding on a logical next step (further testing, design change, stop production, inform a government body, product recall?); especially for safety-related failures. In the case of the Philips medical device product recall, they would be obliged to do a failure analysis by USFDA 21CFR code or the EU MDR regulations, as regulated products like this are legally bound to do so if there are failures that could cause harm.
Ultimately, the results of the FA help you to improve your product validation process for the future and avoid similar mistakes. (09:41)

What failure analysis methodology could you follow?

At Sofeast we use adapted failure analysis methodology coming from tried-and-tested FRACAS (failure reporting analysis and corrective action system) and MIL-STD 785b.

There are 7 steps:

First, decide who will work on this analysis and what the timelines and project context are.

1. Gather information – reviews of samples, documentation, previous analyses and test reports done, customer complaint log, etc.
2. Document any information about the failure – review samples, determine the nature of the failure, do some basic testing, document the manufacturer’s comments about the failure without being influenced by it, and ask more questions to customers to understand if their conditions are a cause (such as the temperature of locations), etc.
3. First analysis – do simple non-destructive testing to use the failed product samples and get an understanding of what happened.
4. Deeper analysis – destructive testing to help you get to the point where you can describe the physics of failure in detail (the chain of reasons that eventually resulted in the failure) without being too optimistic so you focus on the worst-case scenarios. Get as much documentation from the manufacturer and components suppliers as possible to help with testing. Then perform testing for reliability or safety to reproduce the issues by pushing the product to its limits (you may adhere to specific product safety standards for the testing, such as for their flammability) to see how bad the problems can get. Document the results, good or bad, so you have a record of how the product becomes unsafe or whether more testing is needed.
5. Containment, if needed – If a problem is found, what do we need to know and do to contain it? The actions you can take escalate: Test past batches to confirm if there are issues throughout. Stop sales. Give a product warning to customers who have pieces from a certain batch that it may be unsafe and to stop using it until more is known. Start a product recall. Speak with lawyers if a safety issue has caused harm (!).
6. Corrective action plan – devise plan and countermeasures with priorities. Who does what? How can we confirm that the corrective actions (fixes) have worked? Plan ahead to document all of this.
7. Corrective action implementation & follow-up – verify and validate the corrective actions. Document what has and hasn’t worked. If there are new learnings you may go back to earlier steps and rethink assumptions and data that turned out to be incorrect.

Closing the loop is critical. Confirming and validating that issues have been corrected by doing thorough quality, safety, and reliability testing (similar to what was or should have been done during new product development). (19:02)

*****

Have you performed a failure analysis? How did it go? Let me know by commenting, and contact me if you have any questions.