I’m not disputing you, but the snip here raises some questions.
ISTM that most recalls affect vehicle/model years no longer in active production. The statistical discovery and government decision process is so slow that by the time they realize that some particular turboencabulator is defective, it’s 4 model years out of date.
So in that case, the running factories & fresh-built-as-yet-unsold vehicles wouldn’t be involved. Part would only need to go to dealers for eventual refit as customers bothered to bring in their vehicles.
Certainly there can be bad parts batches caught at the factories and in-service problem discoveries that are not (yet) recalls. But unlike a recall, in that situation there’s no unequivocal legal mandate to prevent building and shipping cars with the less-than-ideal parts while replacements are designed, prototyped, tested, their production spooled up, and the pipeline filled.
Clearly there’s a tradeoff here where the corporate legal dept has a balancing act. Minor problems can be dealt with in service, more major ones stop the line. Safety, liability, profitability, and brand image all go into the mix.
With all that intro (and please revise any/all of it as needed) here’s the questions:
What percentage of actual government recalls are swift enough that they affect current production vs. are well after the fact? Of self discovered non-recall problems, how many are severe enough to disrupt production vs. are handled on a forward fit as available basis?
I’m trying to get some feel for the overall shape of the landscape here.
You need a service infrastructure in order to service something. Not a problem if you’re making low numbers of cars. MAJOR problem if you’re producing them in large numbers.
it isn’t even so much about the discovery and engineering analysis taking time, it’s that it could take years in service before a recall-able safety problem shows up. like the rear axle breakage on the Windstar. the problems didn’t even start showing up in notable numbers until long after the thing went out of production.
yes. See: Takata.
I don’t have any numbers, sorry. I can (again) point to the Takata situation as an example; Some companies are continuing to use Takata ammonium nitrate inflators while alternate solutions are being spun up.
but aside from all of that, I don’t know why you all are focusing on recalls. we’re talking about a company which is shoveling incomplete cars out the door then later shipping parts (at what must be considerable expense) to stores to complete assembly. That’s asinine, that is not “the norm,” regardless of what a couple of posters here want to believe, this is a big deal, and talking about how recalls are typically handled in the industry is beside the point.
I’m a latecomer to the thread and have no particular interest in Tesla. If indeed they’re knowingly shipping incomplete vehicles for field completion that sounds utterly unprecedented plus grossly uneconomic and makes me wonder what the hidden reason is. Perhaps a loan covenant or governmental subsidy issue where the rules say they need to “ship X units by Dec 31” to meet some gate and they’re playing games to meet the number X since nobody ever said the “shipped” vehicles had to be, you know, fully saleable as-is.
You were talking about line stoppages and I was curious how often that stuff happened given my anecdotal experience that recalls tend to affect only prior model years. I understand that you were responding to earlier questioners. And that recalls are a very different situation from whatever Tesla is apparently doing.
I understand your answer and we have the same kinds of problems and solutions in my industry. When stuff proves less durable than planned (or downright failure-prone) it’s still used as-is until a replacement can be fielded. Perhaps with more active watching for signs of distress before the actual failure. Even our equivalent of “recalls” generally involve continuing to operate the “defective” parts until they can be replaced in a phased non-disruptive manner.
The exceptions to this gradual approach are so disruptive, and so rare, that they’re headline news not only in the industry, but in the ordinary press as well. With Wall Street duly taking note and tanking the shares of whoever screwed the pooch.
There seem to be several issues being conflated in this discussion.
Tesla is in a unique position for an automaker and in a particular position that they will be in only once - either because they will succeed at becoming a high volume producer and possibly in a disruptive manner, or because they will fail at it and become the Edsel of tomorrow.
The unique position for an automaker bit includes not only that they are an EV but that they are re-inventing aspects of the production and consumer process and bring a tech company product release and update cycle to an industry that has never functioned in that manner before. Will “different” be better, or worse? Obviously posters here have strong thoughts about that. Some of us less strong and will wait and see. But different is not automatically either better or worse.
The unique position for them bit includes that they are in a critical juncture point. trying to build and ramp up into a high volume (and presumptively narrower margin) production (supply, manufacture, to delivery and immediate support) phase is something they will only do once, succeed or fail. Judging how the system will perform when fully launched by how it works as they are building it is silly. And while flawlessly meeting all of their goals would build confidence that they will succeed, stutters in the launch phase are not too surprising and processes that will not work well at full production levels may make sense for a variety of reasons during this unique period of time.
To me the shipping to complete some in the field is part of that unique time. It is, I think, unprecedented, but what they are doing is unprecedented, both for them and for the industry. If it was going to be long term standard operating procedure it would be grossly uneconomic, but if it is awkwardness associated trying to break out of a chrysalis and unfold, so to speak, then those costs are small in the scale of the company’s transition period.
Again, longer term, say a decade from now, I doubt Tesla will succeed in being the disruptive actor in auto manufacture. A moderately successful company with a loyal following maybe, and possibly a (but not the only) leader in vehicles for autonomous driving fleets. But in terms of predicting that arc their doing things differently than others do, a few months behind on target dates, and short term fixes on the fly during the launch build process, just do not seem like “big deal” things that assure their doom to me.
(edit: this was in response to LSLGuy, DSeid posted while I was writing this)
the other thing is that the Takata situation is its own massive pooch-screw on an unprecedented scale. I don’t think there’s been anything which as been anywhere near as big and affected so many car companies in one go. I do remember reading that the affected automakers had to jump through a lot of hoops to be able to talk with each other on how to handle it (anti-trust/collusion concerns.)
But in general, my experiences with assembly plants is that if you’re in regular production the goal is to keep the line moving. A defective part won’t necessarily stop the line; e.g. if they install the dashboard and at a later station power the vehicle and find out that the radio doesn’t work, they don’t stop the line; they just tag the VIN, finish building the vehicle, divert it to the repair bay to have the radio replaced, and re-run it through final. in this case, they’re still building complete vehicles, just one ends up needing a part change and re-inspection before it’s OK To Ship.
A “Stop Build” event is almost always due to either a part shortage or non-conforming parts which can’t physically be installed. At that point it makes the most sense to halt production until the problem is fixed because vehicles could be in various stages of completion and you don’t know if only one part is non-conforming or they all are. Why would you keep building them with missing parts if you have to then have someone take half of the damn vehicle apart to install it once available? So they stop, haul the supplier in to sort and certify good stock, and re-start the line once that’s done.
a “stop ship” is where an issue is found which may affect vehicles which are already built and out in the holding yard. It might be called for because of something found in the plant e.g. a tool out of calibration which wasn’t caught right away, or if a supplier notifies that vehicles may have been built with non-conforming parts. At that point no vehicles can be shipped out to dealers, people have to go out to the holding yards and inspect/tag affected vehicles which then have to be brought back in to the repair stations. any vehicles already shipped, the receiving dealers will get a notice of a “delivery hold” on those vehicles which mean they have to be repaired before they can be sold to a customer.
So yes, it does happen where cars & trucks might need something addressed at the dealer before sale, but it’s not business as usual and God help you if you’re a supplier responsible for a stop build or stop ship. it will cost you dearly.
I don’t think it’s as either-or as that. they may succeed (which is going to take a lot more discipline and realistic goal-setting), they may go bust completely and become the “next Edsel” which I think is unlikely. More likely is that they can’t continue operating as an independent company; they’d need a JV or merger. But their brand is too valuable and they have too much valuable IP for them to just fizzle away.
are they really, though? Some of the stuff I read says that they’re having difficulty with basic stuff like spot welding, which the rest of the industry has been doing for, oh, 100 years. and it 's not like they have to learn how to do it “organically,” there’s plenty of people out there who are either retired from other car companies or looking to make a move who know how to do the “nuts & bolts” of building cars.
but they’re not very good at that either. They got the Model S designed and built because they were able to hire on a lot of experienced professionals who had been laid off or taken buyouts/early retirements from the Detroit 3. but I’m told a lot of those people got burned out and left in a hurry.
Well, I do feel strongly one way. It’s great to do things their way for software-related things such as OTA firmware updates, OTA electronic feature adds, and stuff like advancements in Autopilot. But when it comes to hard parts, you simply cannot expect to make changes every day or every week. Stamping dies are expensive and incredibly time consuming to make (and modify.) You can’t make constant changes to fixtures and tooling. That goes against the very principle of mass production. It’s like the whole “we’re skipping the beta test phase” think Tesla did. Automakers don’t spend millions to billions of dollars on validation testing and prototype vehicles because they like setting money on fire, they do it because you’ve pretty much got to get it right before you build the final production tools. 'cos modifying them after the fact is incredibly expensive and disruptive (not in the good way) and it also means you have cars which are not like each other.
the problem is that there are literally thousands of people out there who know how to do this. Yet Tesla does not appear to want to hire them or listen to them. Probably because they’d say “I’d have to work 110 hours a week? F that S.” (or have hired them and they later quit.)
I don’t think any of this is as “unprecedented” as you seem to think it is. It’s basically accepting Tesla’s explanations at face value.
Thanks for for your last couple of posts. Makes complete sense.
Ref the snip above …
This same criticism has been leveled at Musk’s SpaceX. They chose to “disrupt” a lot further back into the settled production engineering than seems to make sense. And have suffered considerable teething pains there.
The charitable interpretation is that what’s settled is actually old and slow and ripe to be replaced. Existing factories and engineering staffs have legit path-dependent reasons for how they got there, and therefore legit reasons for staying there or evolving only slowly. Reasons that don’t apply to SpaceX’s brand new factories and teams.
The uncharitable interpretation is these guys led with their egos & wallets and simply dismissed hundreds of thousands of man-years of hard earned knowledge as “NIH and therefore stupid. Unlike us.” Cue Hubris, Nemesis, and all that jazz.
So far they’re essentially re-living the entire history of rocket booster development. They’re living it faster than the original process from 1950 did. But they’re aways short of the other guys, and arguably vastly behind where they would have been if they’d chosen to stand on the shoulders of the folks before them rather than obsessively going clean-sheet.
Sounds like Tesla is cut from the same cloth. Which doesn’t surprise in the least.
Not just what you said, but the same criticisms in a much more specific way.
SpaceX also uses a continuous improvement process, to the point where basically each of their rockets is unique. Big bureaucratic organizations like NASA or the Air Force don’t really like this (or didn’t, earlier in SpaceX’s history). They want to certify a rocket and not touch it for years.
Well, that doesn’t work when you’re trying to develop reusability or other technologies at the same time. So SpaceX uses more of a differential certification process so that they can constantly tweak things: adding landing legs, uprating the engines, changing the grid fin design, switching to an autonomous flight termination system, etc.
Fortunately, the Air Force and others seem to have gotten over their hesitation and embraced the style in light of the advantages. The criticisms were never very convincing in the first place, and, like with Tesla, largely amount to “that’s not the way we’ve always done things.”
The NASA engineers thought they were the heroes of the story and they were the only ones who needed to have input on anything. The first astronauts were merely ‘cargo’ in their eyes, did not need to have any instruments or tools on board and didn’t even need a window to look out of.
The similarity I intended to highlight is a corporate culture that rejects industry standards on settled issues such as how to spot weld. And instead prefers to import “talent” from inexperienced outsiders.
Beyond that you’re right there’s not too much similarity; certainly not much in scale.
In some instances, this has paid off; ask anyone about landing and reusing boosters prior to SpaceX coming along, and they would have considered it impossible or so impractical that it would actually *raise the cost.
Citation needed. SpaceX has the bulk of the market share for rocket launches in the U.S. this year. They’re going to have the bulk of the market share for the whole world in 2018:
Starting from scratch is admittedly a stupid idea in a firmly established market such as mass-produced consumer cars.
But doing so in a niche industry where single-use items were running for billions of dollars apiece, was *necessary to get anywhere. Following the mold of other private launch companies would have ran SpaceX directly into the ground. The reason they’ve grown to dominate that market in only 10 years is because they threw established practices out of the window and started from scratch. As it happened, the new processes managed to drastically slash the cost of the business, and so naturally they took that market over.
The thing is (and hopefully this is the last I’ll have to talk about SpaceX in this thread, they’re a completely separate company which has nothing to do with Tesla) is that SpaceX’s biggest advantage is being a private entity. They can do whatever the heck they want, wherever they want to do it, as quickly as they want to do it. NASA, on the other hand, is a government agency and has to derive all of its funding from the .gov. So if NASA ever wants to do anything- say, build a new space launch system to replace the shuttle- in order to get the funding for the program they have to climb mountains of bullshit appeasing this Congressweasel who wants part of their new rocket built in his state/district, another Congresscritter who wants something built in their district, and every bit of new funding they need requires them to kow-tow before a panel of Congressvermin to get it.
plus, the whole culture of “getting things into orbit/space” accepts the risk of catastrophic failures. NASA had tons of rockets blow up before Alan Sheperd made it up there. the Apollo program killed three astronauts before the first mission left the ground. 14 astronauts died in Space Shuttles. And there were at least a few commercial payloads lost when Titan and Delta rockets decided to do a convincing impersonation of a firework. SpaceX has the luxury of doing what they want, including short turn-around times, and being able to work on stuff like landing booster stages after the primary mission (which pays the bills) is complete.
Getting back to Tesla. We’re not talking about an occasional launch of a rocket which is highly stressed and the cost of its failure is baked into the launch price.
we’re talking about a company which has had three disastrous launches in a row- each worse than the previous- finally trying to pull people out of their Accords, Camrys, Fusions, etc. who expect their car to be a trouble-free appliance. It’s one thing to sell Roadsters, Model Ss, and Model Xs to early adopters who just have to have a Tesla and will put up with anything. It’s quite another thing to get someone out of a stone-dead reliable Honda Accord into a Model X whose doors won’t open consistently.
Now, SpaceX has blown a lot of rockets and a lot of deadlines. But, uh, they aren’t behind. They’re now ahead of every other rocket company in all of history. Nobody else has working landings, or rockets made new for an economical cost. And have you seen the carbon fiber tank that’s supposed to be for an upper stage of their massive rocket? Nobody else has one of those - I think the only time one was even attempted was for the X-33 project.
And part of the reason, per Musk, is because they started over. The reason why they had some many oopsies is because most of SpaceX’s parts were clean sheet, not bought from conventional supplies. And that’s because the conventional supplier parts are way too expensive, especially if they need to be customized. Musk explained that the competitors are so risk averse that their engineers are reluctant to use a part in a rocket that hasn’t been flown before - and thus it probably has a 20+ year old design because rocket launch volumes are so low.
This was the only way to achieve SpaceX’s goal of economical rockets - to develop new, cheaper to manufacture parts and go through the teething process for them.
Now, this same strategy probably won’t work in automotive. It might, but probably not. In automotive, an enormous amount of innovation has already happened, innovation the rocket industry skipped, and so even if Tesla *can *do better (machine learning for better factory automation), the cost advantage won’t be very large.
Agree. Tesla has to focus on what CAN be innovated in the EV market and leverage that against the industrial power of his competitors. Musk is at a huge disadvantage with the scale of his automotive ambitions. It’s the equivalent of Ferrari trying to build economy cars in a market that requires the splitting of nickles.
You keep saying variations of this, and frankly it makes no sense for two basic reasons:
Tesla isn’t trying to build economy cars.
Plenty of carmakers build only premium cars and are very successful businesses - BMW and Mercedes for example.
I agree that Tesla needs to cut out the BS and focus on delivering, but it isn’t like Tesla is a patently unreasonable business proposition, unlike the way you talk about.
You travel in a different circle than many of us if you think of that as “economy car.” The Honda Fit starting at about $16K, now that’s a well rated car that hits the edge of being called an economy care by most. The Chevy Spark, the Mitsubishi Mirage … the Tesla 3? No.
A no options Tesla 3 is priced slightly above the middle of the pack of new cars bought in the United States. (Average price new car purchase currently about $34K.) It is of course the entry level Tesla (and the Mercedes-Benz C-class is the entry level Mercedes, and the BMW 3-series the entry level BMW), and yes the whole friggin’ point of the vehicle is that it is to be mass produced in a significant departure from its other models … if they can pull that transition off.
The question will boil down to whether or not the established systems of the majors in the auto manufacturing world have legacy procedures that are inefficiencies today, for these sorts of cars, in today’s and tomorrow’s market, that a fresh slate can avoid (even with teething pains as it gets off the ground), or if Musk is instead trying to re-invent a wheel and will discover that, no the wheel design being round actually is still the best for the job.