When a consumer product fails thermally, customers may get “steamed” and demand their money back.  When the failures are frequent enough that the Consumer Product Safety Commission receives dozens of complaints about “melted plastic” and “first degree burns” a few weeks after the initial launch of the product, they may require the seller to pull the offending product from retail shelves and issue a “safety recall” notice to all consumers.  If you consider a product that is being sold at a rate of 100,000 units per month, it is easy to see how quickly the recall costs could add up.

However, the matter could become even more problematic if the supply chain involves multiple entities (e.g. a product designer, a contract manufacturer, and a marketing entity).  When the recall costs are tallied up, the manufacturer and designer could find themselves in a legal battle to determine whether the thermal failures were caused by “design defects” or “manufacturing defects”.

One particularly challenging aspect of an engineering failure investigation is to understand why only a small percentage of all the shipped products fails prematurely.  By carefully examining the failed units, an engineer may be able to identify the correct failure mode(s), but inspection alone likely will not be sufficient to determine whether the root cause was a bad design or low quality manufacturing.

After the failure mechanism is identified (e.g., loose connection or excessive current draw) the engineer should examine and test a large number of “new-in-box” units to see if there is a correlation between parts that are “out-of-spec” and parts that fail when used normally.   If brand-new parts meet the dimensional and functional requirement of the design, it’s pretty obvious that the design wasn’t adequate to prevent the overheating.  On the other hand, a finding that many of the parts don’t conform to the design dimensions (and other requirements) doesn’t definitively prove that manufacturing defects were the cause of the safety problems.

In a recent investigation of a recalled consumer electronic product, this author discovered that 80% of “new-in-box” samples did not meet the design specification…but less than 3% of the samples failed thermally when first used.  Tellingly, we also found that 9% of the samples were not only “out-of-spec”, but “grossly-out-of-spec” and that each of the samples that failed thermally fell into the “grossly-out-of-spec” category.  (Conversely, none of the 20% of the “in-spec” parts failed when used normally, which provided validation that the design was adequate.)

Using “statistical inference” we concluded that it was virtually impossible (48 chances in a billion) for all of the failed samples to come from the “grossly-out-of-spec” population if only random forces were at play – hence there must be a “causal link” between the “grossly-out-of-spec” condition and the thermal overheating result.  Statistical methods proved extremely helpful in illustrating that the manufacturing “nonconformances” were indeed the “defects” that caused the safety recall!

The purpose of “Investigation Anecdotes” is to inform our readers about the intriguing field of engineering investigations.  We hope you are instructed by this content, and we encourage you to contact us if you seek additional information.