I just read through all these posts for the first time today and I have a few thoughts. The first is did the fact that we, as a country, never fully adopted the metric system have anything to do with this problem? So often when dealing with drawings done in one measurement system, then converting to another, there can be round off errors. Could that have contributed to the pins manufactured being slightly shorter than originally designed, prototyped and tested?
It is not all that uncommon for parts to have minor design changes that do not result in a part number change. You will find spec drawings with a rev A, rev B etc, but unless it is deemed that the older design is a problem, or otherwise unusable, the part number may not change and old stock will be used until depleted. This practice is very common in industries and corporations that practice the “continuous improvement” practices promoted by Toyota and Six Sigma. It is quite possible that in 2006, who knew that a small dimensional change in the length of the detent pin was going to cause deaths on the highway.
How many of the other 15,000+ parts used to make your car could have a tiny dimensional error that could lead to a fatal accident? When you think of critical parts, you generally think of brakes, steering, etc. not ignition. Reminds me of a time when I had a car that was not running and I had it in the base hobby shop while repairing it. Some moron from security wanted to remove my base sticker for not being safe. It had a valid safety inspection sticker but he told me that it couldn’t be safe because it didn’t run. I told him the safest car in the world was one that wasn’t running. I guess that as far back as 1970, he could see this ignition problem coming.
But, how soon did GM know that this could be a problem. The Saturn ignition switch is designed to not lock the steering wheel until the key is removed, and that dates back to 1992. Makes you wonder.
Next question, why isn’t the run position of the ignition switch at the 12 o’clock position instead of the 2 o’clock position? Then the weight of the keys would not be trying to turn it off.
The strange thing is, there was no “dimensional error” in the switch manufacturing process. Delphi made it as designed. And they even told GM that the switch did not meet GM’s own torque specs. But GM had them go ahead and manufacture them anyway.
So GM knew right from the start that the switch did not meet their own specs.
How soon did they know it could be a problem? Almost right away, I think in 2001 or 2002 they started getting complaints and reports of crashes.
Good question about the switch position. You would think that’s the obvious thing to do, putting the “run” positon at 12 o’'clock.
@keith I don’t think the metric has anything to do with the switch problem. GM divisions have been in business for about 100 years, They started when the “lowly” Japanese were babes in the woods when it came to quality. When I was a kid, a Japanese bike was a joke; English and American bikes were the best.
They started with quality control in the post war period, coached by Deming, an American quality expert. Those companies realized that quality would sell and gradually improved to where by the mid 80s they were on par with the best in Detroit. The rest is history.
Japanese manufacturers focused on both quality, durability at the lowest cost. GM, ruled by bean counters, focused on “good enough” at the lowest cost. I took a course from General Electric in the 60s called “Value Engineering”. The main focus was on immediate cost reduction, not increasing quality. Honda and Toyota practice six sigma quality control in manufacturing, but the real value is in their long term durability testing and improving instrinsic quality and durability.
I used to work for a company that cast cylinder heads for the big three. There were a lot of drawings that were converted back and forth between metric and SAE. The company I worked for was an Italian owned company that had factories in the US and many other countries all around the world.
Many parts are overdesigned so that no failure ever occurs. This allows for some errors in design or manufacturing as long as they are not too severe. This tendency to overdesigned my be why GM was willing to use the parts even though they fell short in the torque specification. Clearly, the wrong decision in this case.
Still, regardless or who knew what when, shoulda woulda coulda, and on and on, thats 2.6 million cars with 13 fatal accidents. Maybe my $2 Japanese calculator is wrong but that’s a .000005 failure rate. Take out operator error and I don’t know what it would be. Of course any failure rate can be improved on but even Deming would admit you aren’t going to get to zero without redundant systems that would make the cost of the product not fit for the purpose intended.
The important thing to remember is that it’s all about perception by the consumer. If GM sales are strong , it will be soon forgotten. If sales plumit, blame will have to coincide with some " reason" and this seems as good as another. So, how many Toyota executives were fired in the " unintended acceleration" scandal if sales remained strong ? CEOs and upper management will then me praised for strong leadership during tough times and not being responsible for consumers dying. Having accidents attributed to car malfunctions is an acceptable part of the calculus. Being blamed for it and having sales plummet, is not.
@Caddyman
How many is not as important as whom it affects. A few people who loose a grandchild or relative will still think those “astronomically” small numbers are unacceptable. The rest of us will listen if those involved have the means to make us listen.
@Bing, the actual failure rate of the switch is much higher. Many more complaints have been filed about the switch accidentally shutting off in dangerous situations that didn’t lead to crashes. And probably for every one non-fatal incident reported to NHTSA, there are many others that people don’t bother to report. Furthermore, I suspect those 13 fatal crashes are just the tip of the iceberg.
According to analysis described in the article below, almost 1200 people have died in crashes of the recalled vehicles in which the airbags did not deploy. Of those nearly 1200 deaths, 303 of them were front seat passengers who did not get the benefit of airbag deployment. Why didn’t the airbags deploy? Quite possibly because the ignition switch was in the “accessory” position at the time of the crash. And did those crashes occur in part because the ignition switch accidentally turned to “accessory?” Unfortunately, not enough data was collected on those crashes to nail down why the airbags didn’t deploy or what caused those crashes. If that data had been collected, it’s quite possible that we’d see this ignition switch implicated in hundreds of deaths, not just a handful.
Airlines are ultimately prepared to accept a plane auguring in now and then with 200 people on board as the cost of doing business so why wouldn’t a car company do the same.
From the New York Times hot link there is a new excuse for failure to fasten the seatbelt as follows. Quote: “In an interview on Wednesday, Ms. Chansuthus’s brother David said the family was troubled by the air bag’s failure and the fact that his sister had hit her head on the steering wheel, which he said, suggested that the seatbelt had failed to lock.” Unquote.
Airplanes have redundant systems to keep the airplane flying if there is a failure. Travel in a car is an inherently dangerous activity and redundant crash protection, seatbelts and airbags are good to have.
Now reports are coming in saying that the ignition switch is prone to cutting off even when there are no other keys hanging from the keyring. Just hitting a bump in the road caused this guy’s ignition to cut out in the middle of a curve, and he spun out and crashed.
"The National Highway Traffic Safety Administration has received more than 100 complaints about cars with the ignition defect since G.M. initially notified owners of the recall in early March. The filings, concerning episodes that had taken place since March 10, included complaints from people who were frustrated that they could not get their cars fixed right away, but also from those who described unexpected stalling, steering failures and bizarre situations, like the car continuing to run after the key had been removed from the ignition.
“My car has been turning off/stalling in the middle of the middle of heavy traffic for the past three weeks,” one Cobalt driver wrote on March 13.
“We’ve been getting many, many reports of folks that have been having problems with their ignition switches regardless of what was on their key chains,” said Eugene R. Egdorf, a lawyer with the Lanier Law Firm in Houston."
One needs to be wary of thinking of the New York Times as an objective source of news of an engineering nature. They mix facts with entertaining conjecture from lawyers and from others who also might smell money in the chaos.
Bloomberg has identified the engineers who were suspended:
As expected, one of them is Ray DeGiorgio, who quietly changed the switch design and later denied it under oath.
And here’s a quote from the article with some interesting insight on the part number issue:
"The redesigned part was assigned the same number. That is unusual, according to auto-safety experts, because the automaker, as well as dealers, repair shops and auto-parts stores, would have no way to tell the updated part from the old.
“It’s pretty much standard procedure to make a part number change when they change the part like they did,” said Pat Donahue, a private engineering consultant who worked at GM for almost two decades until 2001. Typically, he said, such a change would also require authorization from a manager. DeGiorgio was the only person listed on his request."
"By 2012, GM knew a switch flaw was causing some Cobalts and other models to stall and prevent airbags from deploying, according to Lance Cooper, the Georgia lawyer who deposed DeGiorgio and several other GM engineers as part of the wrongful death suit against the automaker. Yet GM’s inhouse investigators, who had been looking into the flaw for two years, couldn’t figure out why it wasn’t showing up in later models, Cooper said.
“In part, that was because DeGiorgio didn’t tell the investigators he had ordered the redesigned part without assigning it a new number, Cooper said last week in an interview at his office in Marietta.”
It doesn’t look good for Ray DeGiorgio. But one of the reader’s comments on the Bloomberg article I posted above had an interesting theory that DeGiorgio was trying to do the right thing despite management’s refusal to spend the money necessary for a “real” redesign of the switch:
Comment from “alonsoperez”
“It appears to me that DeGiorgio hid the change because he could not get approval from management (apparently Altman), because there was no “business case”. I’ve seen this before, where lower level people try to do the right thing despite instructions from management. If so, DeGiorgio is the hero here, as his action almost certainly saved lives. If you ever drove a 2007 or newer Cobalt, you should thank him.”
Yep that sounds a lot more plausable. That’s the way organizations work sometimes. Just have to watch out for underlings that tattle up the ladder to make themselves look good.
The really interesting question now is what does GM do with the Cobalt engineering manager Altman who nixed the proposed fixes because they would cost a measly half million dollars.
As opposed to the billion plus that the recall and lawsuits will cost.