revised version posted by Smitka at The Truth About Cars
The Takata airbag inflator problems illustrate a fine dilemma: quality standards across the auto industry are good, those for safety-critical devices very good. The result is that only things that occur very rarely get through the production process, and many of those either cause no problem or don't get reported. That makes confirming that there is in fact a pattern challenging, and figuring out the root cause (or causes) extraordinarily so. The number of known deaths (the media suggests 5) is a very small fraction of the number of lives saved by Takata airbags. So the other dilemma is that the fundamental robustness of the manufacturing process means the benefits of a recall are also very low, and the quality of work in your local dealer's repair bay is not equal to that in an airbag plant or vehicle final assembly plant. The cure can be worse than the disease.
Anyway, I spent a day this week with an airbag manufacturer, listening to engineering presentations on a new airbag design from the supplier's senior engineers, with a senior car company airbag engineer also in attendance. I won't name the companies, and what I write is based on information that should be available from public sources. (I spot-checked a couple of the points.)
1. First, globally there are tens of millions of Takata airbags on the road. I've not been able to find a number, but I would guess that over roughly 1 million such vehicles have been involved in a collision that led to an airbag deployment. Of those, to date there are 5 known fatalities and several more injuries. Actual problems are exceedingly rare.
2. The cause is as of yet unknown, as there are multiple failure modes. Which one(s) are leading to the observed problems? Small numbers mean (i) this analysis is intrinsically very challenging. It is complicated by (ii) the evidence going up in smoke when an airbag inflator explodes and (iii) other evidence going up in smoke because documents in Japan were sent to the incinerator. The concentration of incidents in very high humidity locales suggests deterioration of the ammonium nitrate "propellant" due to hydration, wich could cause the sheets of material to turn into clumps (sheets go "whoosh," clumps go "boom"). However there are several incidents in areas not known for high humidity. So there could be two different problems, or one systematic problem and the random one-in-a-million manufacturing defect, or all random problems some of which just happened to be clustered geographically. So last week automakers who use Takata airbags got together to decide how to jointly collect and analyze disparate data in the hopes that the combined data would allow meaningful analysis. This was a meeting cleared by the Dept of Justice as not violating antitrust because it was limited to engineers discussing a narrow set of issues. Almost every car company uses Takata for at least a few airbag applications, so it was basically a meeting of the global customer-side engineering community.
3a. If the actual problem is not systematic, then a recall may do nothing at all except cost lots of money, because the same one-in-a-million bad inflator ratio won't change. If anything, a rush to increase production will make monitoring production process compliance more challenging and could lead to a higher number of (idiosyncratic) random defective airbag inflators in cars.
3b. Again, other manufacturers cannot substitute their inflators for a Takata inflator -- they would have to design a product that matched the gas generation profile needed to match the Takata airbag, verify their method of manufacturing produced parts that actually worked to design, test prototypes with the Takata airbag to make sure there was no unforseen interaction (vent angles or orientation slightly different, lots of subtle interactions). Then they would have to set up a production facility, run off a lot of parts coming through the actual production process on the machines and tooling and inspection processes that would be used (rather than the prototype build process), and have these tested and retested. This is necessary because the bag portion is very, very specific (the exact grade of material and how it is folded are all very carefully specified, tested and then monitored during production for exact replication). It would be very hard to do this in under 6 months, and production does not ramp up from nil to full overnight. It would be impossible to do this in 6 months across all of Takata's airbag-inflator-vehicle combinations, because each would need to be tested separately. Engineers can work 16 hour days for a while, but not for month after month. There isn't excess engineering and testing capacity just waiting for a recall to come along, and car companies want their engineers to continue working on new vehicles, they don't want to stop everything under development to re-engineer an old (perhaps decade-old) product.
If you need to find a needle in a haystack, maybe it's not worth finding the needle.
4. Nevertheless, of the inflator manufacturers, as far as I can tell Takata is the only one whose inflator operations did not start out as a division or factory of a rocket engine or explosives company. Instead Takata was a cut-and-sew operation that had expertise in fabrics that then added in-house pyrotechnic capabilities. That adds to the suspicion of a systematic albeit very rare propellant problem, but again, the number of incidents remains very small and there is essentially no ability to cull the necessary information from incident reports or (when they were kept) piles of shrapnel.
5. For reference, manufacturers of inflators include Autoliv (the other really big player), TRW, Key Safety Systems, Daicel and (making only inflators) ARC.
6. Finally, I want to reiterate that the numbers indicate you are much safer in a car with a potentially defective Takata airbag than a car without any airbag. The Takata airbag defect matters only in a frontal collision. Even if the inflator did spin off shrapnel, which is (order of magnitude) perhaps a 10 in a million chance, the chance you will be seriously hurt is lower. If you don't have an airbag, you'll be using your head -- to slow down the rest of your body. That story never has a good ending, and can readily have a fatal one, the latter at a rate much higher than 10 in a million.
mike smitka, from Toronto