Shoutout to the Henry Ford Museum located in Dearborn, Michigan which is a treasure trove of 19th and early 20th century technology. A true celebration of, what I would consider, the most physically transformative era in human history thus far.
Eventually it was even cheaper to not have a separate crankcase and just cast it with the rest of the engine block, as is now common practice (and has been for around a century).
- https://en.m.wikipedia.org/wiki/Henry_Ford#Antisemitism_and_...
A friend of mine ordered a set of dimensioned drawings for the P-51 Mustang. He was investigating the possibility of going into business making P-51s.
But when talking to people who owned P-51s, he was told that the drawings were made after-the-fact. The true design was encapsulated in the jigs and machinery developed for the factory floor, and they'd all been scrapped after the war.
The drawings were useless.
I'm pretty sure that P-51s are maintained these days by making replacement parts by hand and custom-fitted, a very expensive proposition.
I'm waiting for history to repeat itself with Tesla, but it's not a popular (hi)story to tell. Not as popular as how great an American pioneer Henry Ford was, for sure.
And I thought it was Toyota which pioneered the 'just in time' method.
The real dynamic was that efficiency and scale compound improvements but also compound lock-in. The more Ford optimized his system around one product, the higher the switching cost to change anything fundamental. Every special-purpose machine tool, every supplier contract, every material flow was tuned to one car. At small scale, that’s agility. At massive scale, it’s a straitjacket.
Tesla faces a version of this trap. Its efficiency engine is vertical integration and battery/powertrain mastery. But the stronger that engine gets, the more risk that its identity collapses into “this is what we make, as efficiently as possible,” rather than “this is what the market wants, however we must adapt.” GM in the 1920s wasn’t just adding variety for fun, it was creating a systematic upgrade ladder (“a car for every purse and purpose” as Sloan said at the time) that turned consumer churn into a growth engine (allowing customers to start with basic models like Chevrolet and progressively upgrade to more luxurious brands such as Oldsmobile, Buick, or Cadillac). I agree that Tesla hasn’t yet built an equivalent mechanism to capture customers once they’ve “had enough of the Model T.”
The irony is that efficiency-driven firms almost never stumble because they stop improving; they stumble because all their improvements are local optimizations. Ford’s engineers in 1925 were still making operations faster, parts cheaper, and tolerances tighter, but all within the cage of the Model T. Tesla today is in danger of repeating this exact logic trap: world-class at batteries and drivetrains, but perhaps blind to the fact that consumer perception, design novelty, and product line evolution can erode even the strongest cost advantage.
As long as parts were mechanically produced in series, they are standard. It doesn’t really matter then if the drawings were made ex post from existing parts. If you produce from a description with a precision equivalent to the original machinery tolerance, you should end up with virtually identical parts.
> I'm pretty sure that P-51s are maintained these days by making replacement parts by hand
This is fairly common for old cars too when spare parts are not mass produced anymore if they were at all. It used to be really common for all repairs not that long ago to be honest.
My grandfather was a mechanics in the 50s and he liked to explain that machining parts was a common occurrence then because getting parts shipped would take considerably longer than just making them and everyone involved knew how to machine parts anyway.
You don't. Tolerances are almost never bespoke except in the rarest occasions of cutting edge green field development.
You can literally buy an old book (the info is all on the internet too but not in one place) and skim through thousands and thousands of pages that tell you what different classes of tolerance for different applications are. You don't need to know what they spec'd. You just need to know what the part was and how it was used.
The thought process goes like "This worn the f out coupling is .9443 inches major diameter, so it was a 1" (nominal) coupling, and it must slip under load, so I will use a class-whatever fit, and the tolerances for that are +/-.0gfy"
An imprecise drawing is also a sufficient starting point for the process.
And like 30% of the time then you get half way through figuring out how to set it up and realize that you can literally just buy whatever you're looking for at tractor supply or Alibaba or whatever and modify it slightly.
If you farmed the drawings out to machinists to make the parts, they would not fit together.
One of the things I did in my engineering work at Boeing was to work out the tolerances for each part. The idea was that if the parts were within tolerance, they'd all fit together. If the tolerances are missing or incorrect, then you're in for a lot of very expensive rework.
Some of the engineers were incompetent, and when the first run of parts were received, they could not be fit together. I'd be assigned to fix it. The assemblies I worked on all fit together first try.
The stab trim system had very powerful motors, and if the jackscrew nut was run into the stops too many times, it would tear it apart. So there was a mechanical cutoff system with cables and levers which would turn the motors off between 1/4 and 3/4 turn from the stops. It was not adjustable, since Boeing did not trust mechanics to adjust it properly. So, it relied on accurately machining the parts. I got the job of figuring out all the tolerances.
On the first build, I got a call from the factory saying, since I'd designed it, I get to be there for the first rigging. I arrived, and the seasoned mechanics laughed at me (I was very young) and said the rigging never worked the first time, and would have to be redesigned. We all crawled into the fuselage behind the rear pressure bulkhead. They manually set the cables to have it 1/2 turn from the bottom stop. Then they chortled, said they were going to break it, and turned the motors on. (Those motors make quite a racket.) They watched the stab trim go full travel to the top. The cutoff system shut off at exactly 1/2 turn from the end, dead center in the tolerance zone. I said "bye guys", climbed out, and went back to my desk.
(Yes, I know about wraps, substitute "purple" for whatever feature or body style or quirk it is you have trouble finding on modern cars)
Maybe Rivian will fill this niche if Tesla doesn't.