Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Machine that makes the machines

This site may earn commission on affiliate links.

Cobbler

Paranoid T.E.S.L.A Bull
Sep 22, 2015
716
9,814
België
From MP2:
What really matters to accelerate a sustainable future is being able to scale up production volume as quickly as possible. That is why Tesla engineering has transitioned to focus heavily on designing the machine that makes the machine -- turning the factory itself into a product. A first principles physics analysis of automotive production suggests that somewhere between a 5 to 10 fold improvement is achievable by version 3 on a roughly 2 year iteration cycle. The first Model 3 factory machine should be thought of as version 0.5, with version 1.0 probably in 2018.

Can someone explain this in further details? What does Elon mean by this?
How could this differentiate Tesla from other big manufacturers? Is this just a reference to a much closer cooperation with an external supplier?

I have a basic knowledge about assembly lines, robots and networking (PLC engineer). I've seen the YT-video's about the Freemont Tesla factory and, to be honest, don't see a lot of difference in how other companies make cars. Tesla is relying on external parties for their automation, the same as other companies. I know that robot manufactures sometimes create customized product-lines for their big customers.

Apart from the car production, I've already read on these forums that Panasonic worked on new machines for their battery production in the Tesla Gigafactory that will be are faster and more efficient to do the job. So if Elon would be talking about battery production instead of car production, it would've made more sense to me.
 
I too find that paragraph intriguing and possible the most interesting from the SMP2.

My take is that Elon bets on optimizing the *sugar* out of everything he touches (being a machine, being a person). My guess he sees Fremont not even near as optimized (speed and cost wise) as SpaceX.
Considered this last one and the first approach on building a rocket:
Elon knew a rocket cost X dollars to build from the ground up. But what was the cost of the sum of all the constituents of a rocket (iron, carbon fibre, titanium, etc etc)? The answer: 10 times less.
Is Fremont at that stage? I doubt. Is the Gigafactory there? I bet it's close.

So my guess Elon whats Tesla factories the reach that kind of optimization. And create procedures that allow creating various interactions of the same factory around the world, as far as external providers allow it.
 
Sure. First facts: it is hard to tell is he is talking about volumetric efficiency or not when he uses 5 or 10 fold improvements. Listen to the shareholders meeting. Or the conference he went to at about the same time. I think the basis for this statement in context now would be Tesla production today. So 5 to 10 times better than what Tesla is doing today is how I would interpret that.

Now opinion: The "2 year iteration cycle" says he is listening to the wrong people. Maybe people who are getting ready to sell him an excuse. He has made time linear, or serial, rather than parallel. This slows things down a lot. Maybe the all nighter made him into a fatigued coward? (That is either a Lombardi or Patton quote, and is in no way pejorative). Improvement needs to be parallel, at the very least "Tick, Tock" if he does not want to risk running out of money. 3 lines where improvements are rippling through continuously should double the improvement rate over that obtained with "2 year iteration cycles."
 
He carefully didn't say anything about the assumptions for the iteration cycle, he just said where he thought they could get by version 3 with a roughly 2 year cycle, starting in 2018.

That's to a 5-10 fold improvement. Pretty big deal.

2018 he expects the Fremont plant to be approaching it's traditional maximum capacity at several hundred thousand vehicles.

So 6 years after that around 2025 he expects Fremont to be able to produce 1.5 to 2M vehicles... 4x as many as it ever did under GM or Toyota and a lot more than any assembly plant has ever produced.

He seems to have gotten focused on designing vehicles for production and how much difference that can make. Tesla already has a huge potential advantage with an architecture that's extremely modular and simple compared to traditional ICE cars.

One big change that could help with this is going to full drive by wire for M3 at first and further along with all models.

If M3 is built hardware wise to be fully L4 autonomous it really doesn't need physical linkages for steering, they serve no purpose in a car designed to operate without anyone in the driver's seat.

If they can go drive by wire the vehicles get quite a bit easier to manufacture. I think of the interior of the M3 demo as exactly what to expect. No drivers gauges built in, a symmetric dash. Plug in steering wheel, pedals and Heads up display allow any car to be left hand, right hand or no-driver configuration.
 
So 6 years after that around 2025 he expects Fremont to be able to produce 1.5 to 2M vehicles... 4x as many as it ever did under GM or Toyota...
I would not assume that Elon's projections about factory efficiency apply to the Fremont plant, which was built decades ago. I believe his projections apply to future plants that Tesla will build from scratch.
 
Can someone explain this in further details? What does Elon mean by this?

I've seen the YT-video's about the Freemont Tesla factory and, to be honest, I don't see a lot of difference in how other companies make cars

You are right, what he is saying that they are going to put their engineering resources into designing a better factory, that version 1 of the factory that they are working on will be running in 2018, with a 5-10 fold improvement in cost/speed/quality achievable by version 3 of the factory in ~2022.

Looking at another industry... Samsung & GE have productized large molecule pharmaceutical manufacturing... with GE taking the approach of selling the factory components to pharma companies and Samsung using the factories to sell services as a contract manufacturer; both have productized the factory, but are taking different go to market models.
 
I've wondered about 'factories that make factories' too and come up with 'factories within factories'
I come up with aspects of vertical integration unrelated to marketing such as
  1. Shared expertise. This might be the most important since EM has at his command an immense brain trust.
  2. Shared product development. E.g., GF designs a battery to best match EV and PV
  3. Proximity
  4. Shared materials
The Japanese have highly integrated and tightly controlled supplier networks that share some of these ideas.
 
If M3 is built hardware wise to be fully L4 autonomous it really doesn't need physical linkages for steering, they serve no purpose in a car designed to operate without anyone in the driver's seat.

If they can go drive by wire the vehicles get quite a bit easier to manufacture. I think of the interior of the M3 demo as exactly what to expect. No drivers gauges built in, a symmetric dash. Plug in steering wheel, pedals and Heads up display allow any car to be left hand, right hand or no-driver configuration.


The 3 is a transitional vehicle. Expect real connections electric assist.
 
I would not assume that Elon's projections about factory efficiency apply to the Fremont plant, which was built decades ago. I believe his projections apply to future plants that Tesla will build from scratch.

The comments I've seen suggest that he's talking about Fremont at least for the 3 iteration cycles. The fact that it was built by GM a long time ago doesn't necessarily interfere. It's just a shell. The actual machinery and internal processes are mostly new.

If Elon hadn't come to the conclusion that Tesla could reinvent the manufacturing process at Fremont and greatly increase production well beyond what the plant ever did under GM or Toyota, he'd have had to start construction of a new plant now to meet his target of a million vehicles by 2020. It actually would be too late to even have those targets if a new Tesla plant was needed.

I think it's also implied that once they reinvent the manufacturing process, the machine that makes the machine, they will roll it out to other new plants built from scratch. That's similar to "Gigafactory 1". That one plant will be enough to handle all their battery needs for a million cars plus microgrids in 2020, but after that it would be replicated in new locations.
 
The 3 is a transitional vehicle. Expect real connections electric assist.
I realize it's a long shot that bumps up against current regulations. It is an good example of a dramatic design change that enhances manufacturability though. The traditional auto industry has been very timid about it while working toward it for a long time.

I was curious about the meaning of Elon's comment about the steering of the M3 demo that it wasn't the actual version...that the actual version would be IIRC 'like a spaceship'.

If M3 is going to be designed and built equipt for full autonomy (which might be too ambitious as well) then full drive by wire makes a lot of sense. Getting there in an M3 built next year or 2018 is a pretty big stretch though.
 
The proposed ramp-up is extraordinary.

-1x-1.png


The real advance with Model 3 must be the new approach to the production line.

(Or rather lines)
 
It appears to me that Elon is driving towards a production line that is fully automated. No humans. That is the only way you can make a line move much faster than it currently does, and Elon has spoken many times about how slow current production lines operate.

So how do you make a fully automated production line? One way would be to design the vehicle so that all the components requiring human hands to assemble be components assembled in advance, away from the production line. They are then staged at the production line and robots insert and attach them to the vehicle body in white. That way the line can move very fast.

That means a fundamentally different design of the vehicle, optimized for automated assembly. But to the customer it won't look any different.

There will still be humans involved in production, just not on the line.
 
  • Informative
Reactions: croman