This is a good piece on Tesla’s coming “unboxed” assembly process, which they introduced at Investor Day earlier this year.
-IGU-
7 Likes
I worked in the car biz for quite a long time and certainly agree there are many opportunities to improve.
And, I’m sure Tesla has proven… in CAD… that this new process works just fine. But…(from the article)…
“However, there are some quality-related risks involved, such as potential gaps in fit and finish,” warns Pischalnikov.
It seems there are very few manufacturing experts posting at the Fool. At least I seldom see such an animal running around the boards. But… for any doubters… there is an immense gap between what is proved on paper (or on the screen)… and actually implementing it in reality… day in, day out… thousands of times… tens of thousands of times. It’s really hard!
I suspect there will be a LOT of “hands on” development… in the factory… to be done to ensure everything fits together and looks good, especially on the interior. Because CAD representations of parts are not quite the same thing as parts in reality, especially for floppy stuff like carpets and other interior trim. It can assemble brilliantly looking at the screen, but there are likely to be issues to work out when the actual parts come in the door.
I certainly hope they can get those details worked out easily and quickly because any fundamental error in the concept could be a disaster… and since TSLA has a significant part of my portfolio… I want continued success in lieu of disaster.
As I write this, I’m re-assured by the ASSUMPTION (on my part) that they’ve probably already been doing prototype trials to prove out some of the trickier aspects. At least I HOPE so! 
Rob
He is no fool who gives what he cannot keep to gain what he cannot lose.
3 Likes
I’m no manufacturing expert but I do watch videos by manufacturing experts, namely Sandy Munro. At one time he pointed out how cars on the assembly line had to forced (with crow bars?) to make them fit properly. At another time he pointed out how giga castings improved accuracy by eliminating the need to weld, glue, bolt, or otherwise join together hundreds of parts.
When Pischalnikov made his comment, did he take giga castings into consideration or was he talking about the traditional assembly line? Musk has said that prototyping is easy while manufacturing is hard. I doubt that the Monterrey plant is pie in the sky. Problems will be solved, maybe on Elon time, but solved. 
If it were someone else I might worry but Musk is the father of the Model Y which has displaced the Toyota Corolla as the best selling car in the world. Quite a feat. What does Pischalnikov say about that?
The Captain
Edit:
BTW, this will not be the first time that Tesla tinkers with the assembly line. At least in Texas and probably in Berlin Tesla is using a cell system where cells work in parallel and any cell can try innovations. Once the innovation is tested and quality checked it goes back online and the other cells incorporate the innovation.
BTW2, while working as a management consultant I had the opportunity to visit lots of manufacturing plants in part because we never took what we were told as fact until we checked it out in the field.
3 Likes
That, I think, misses my point and perhaps his point.
I’m not so much concerned about dimensional tolerance and stability of large castings. I mentioned concern about floppy parts… using carpet as an example because carpet has laughable dimensional control… but it goes beyond that.
The interior of the car… has a multitude of parts that overlay each other (remember that word “overlay”) and have high variation in their shape, stiffness and stability. It will be a challenging process to design completed sections of the vehicle to just “bolt together” and to ensure the process of overlaying adjoining sections being assembled achieve a pleasing fit and finish. And over all that… what you see on the computer screen will NOT be what you’re putting together in reality because all these floppy parts (and other soft interior parts) are unique and have unique interfaces with each other from car to car. The carpet (a good example) in car #1 will have a significantly different shape from car #2 and #2’s carpet will be unique from #3 and so on. Difficult to just “slap them together” with all the other parts and have it look good.
In addition, each car section being installed together will follow a certain specified path as they mate. Naturally, all the parts will be designed with that assembly path in mind if/as required. And… it will be a challenge… resulting basically in a trade secret… to transform that computer aided assembly process into a process that actually works reliably for each individual car… with it’s own unique set of unique parts. Partly because parts want their OWN path to assemble to other parts because Part A in Car #1 will assemble to Part B in Car #1 differently than most other cars. And there is a large number of these part to part interfaces that have to successfully merge simultaneously as you put together major sub-assemblies.
Because big castings are great… and a huge benefit… not everything is so dimensionally consistent, rigid and stable.
It may be of benefit to consider this analogy: You’ve got a number of plates with spaghetti on them… and you’re wanting to produce… consistently… a 3D shape like a gingerbread house with these plates… floor, walls, ceiling, roof AND…adjoining rooms. And you want the spaghetti on each plate to stay in place while you do this… and have the spaghetti noodles all mesh together in a certain way… from each plate… together.
Now do that a thousand times each day. Every day.
It’s not impossible, but “it ain’t easy” either.
Rob
He is no fool who gives what he cannot keep to gain what he cannot lose.
3 Likes
Addendum:
Upon further reflection, they could simplify this considerably by having some parts installed AFTER assembling the floor/roof/sides/front/rear. That would enable you to cover these problematic interfaces… and you probably need to make final electrical connections anyway.
So… bolt/glue together, make final connections, slap on covers that would likely be there anyway.
Rob
He is no fool who gives what he cannot keep to gain what he cannot lose.
2 Likes
One of the most difficult parts to install are the floppy wire harnesses. Since Tesla plans to have the Optimist robots do it, they are redesigning the harnesses to make it possible for the robots to do it. I think you are losing sleep over issues that Tesla has already considered.
The Captain
4 Likes
An interesting article. I was curious about one point, which seemed to me to be a major omission by the author.
Towards the beginning of the article, they quote an engineer who describes the traditional method of auto assembly - and then explains why that method has been used:
“The reason that’s always been done is for color consistency, to ensure that there’s a perfect match between the doors and the rest of the car body,” Prasad points out. “By not having to assemble, disassemble and reassemble vehicles, you can reduce production costs and eliminate waste.
So, okay - you could reduce production costs and eliminate wastes if you don’t go through the assemble-disassemble-reassemble process. And Tesla’s proposing to do that. But the article never really discusses how Tesla plans to address the reason that this method exists in the first place - to ensure color consistency. This isn’t a manufacturing process that is followed simply because of inertia. There is a problem that the existing manufacturing process solves. That problem is (apparently) the reason why companies don’t do more 2D and less 3D stuff - you get inconsistent paint colors if you don’t do all the painting at once.
Given that, it’s weird that the author didn’t discuss with the engineers how they though Tesla might be trying to solve that.
1 Like
That’s an easy one! Paint all the body parts of each vehicle at the same time. Tag them with IoT chips by paint batch.
The Captain
3 Likes
That’s why the front and rear bumpers of most cars (regardless of manufacturer) don’t quite match the rest of the vehicle. They’re made of a plastic material that can’t stand the heat of the paint ovens used by the rest of the body.
Plus, the manufacturers don’t want to spend the money to make them match.
Rob
He is no fool who gives what he cannot keep to gain what he cannot lose.
3 Likes
But is that what Tesla is planning to do? And would that work? There’s a extra effort and time involved with assembling/disassembling the doors on and off the car. But there’s also effort and time involved with having to match the exact doors to go to which body - because then a mistake also results in delay on the line.
1 Like
I have no idea and no inside info. It’s just that the naysayers keep coming up with negative scenarios so why not come up with positive ones just to bug them? 
All in all Elon Musk has not done too badly, don’t you think?
The Captain
3 Likes
Every assembly line with good process that I’ve seen relies on continuous improvement. The first 100 off the line will only be 90% accurate, and those can’t be sold to end customers. They use them to learn from, and sometimes maybe run them through parts of the assembly line again to fix the problems. The next 1,000 are 95% accurate, maybe sellable, but many customers may not be happy with them. Company will do repairs/replacement as needed. The next 10,000 will have small issues but will be 98% accurate and most customers will be happy with it. The next 100,000 will occasionally have small issues, but by and large customers will be happy with it. The next few 1,000,000 should all come out identical with only rare issues related to tolerances or unexpected changes in a subassembly from a subcontractor. Also, as production continues, there are constant cost reductions (and regrettably sometimes increases) or subassembly swaps as they become available and/or necessary.
Then a few years later, they refresh the model and it all starts over, usually without the first 100 or first 1,000 types of issues (but sometimes yes depending on the extent of the refresh).
2 Likes
Absolutely. The Model Y will be (probably) the best-selling car in the world this year.
But past performance is no guarantee of future results. Building a very successful car company doesn’t necessarily mean that Musk (or the many other engineers at Tesla) have found a solution to the very specific problem (paint consistency) that appears to have underlay the aspect of the manufacturing process they’re going to change.
Tesla’s “clean sheet of paper” approach can give them a lot of flexibility in getting rid of (or improving on) manufacturing processes that exist for no reason other than inertia. But if there’s a real, legitimate problem that a cumbersome manufacturing process exists to solve, then changing the process will require coming up with a solution to that problem. Being a flexible start-up gives you the advantage of being able to slay sacred cows and gore oxen and avoid being hidebound (and any other cattle-related metaphors for being set in one’s ways) - but it doesn’t give you carte blanche to change processes that are actually necessitated by some real, legitimate issue.
3 Likes
I guess it isn’t discussed because it’s too obvious. The various pieces that are to be painted go through the paint shop together, ensuring color consistency, but they aren’t assembled into a single unit until later.
-IGU-
3 Likes
Exactly what I said above…
That’s an easy one! Paint all the body parts of each vehicle at the same time. Tag them with IoT chips by paint batch.
The Captain
1 Like
That’s what captain mentioned. But if that’s the solution, how could there ever have been a problem in the first place? Manufacturers could have been doing that at any time over the last hundred years. The assemble/disassemble/reassemble process is so cumbersome that it never would have been adopted in the first place if you could obtain color consistency efficiently by this method.
Yes, but that’s not how innovation happens. The assembly line worked really well until someone came up with a better idea and had the power to implement it.
The Captain
2 Likes
Yeah, but that doesn’t explain how this situation arose in the first place. If the solution is so obvious and easy that it could have been implemented at any time over the last hundred years, how did every car assembly line get organized around the cumbersome workaround? No one ever would have tried the assemble/disassemble/reassemble route in the first place.
There must be some logistical problems that would be caused by trying to paint all the painted parts as separate pieces at the same time in the paint shop, and then trying to bring all of those pieces together (without error or misplacing) back onto the assembly line - else no one would have ever done it any other way in the first place.
1 Like
Lots of things are simple when looking from afar. “Doing” provides a different perspective.
Most folks at the Fool are, naturally, familiar with a few things and unfamiliar with most. Doesn’t stop from “knowing” about the unfamiliar stuff.
I have a BIT of exposure to the world of OEM painting after spending some time in an automotive production painting environment. It is fiendishly difficult (and messy… lost a lot of ties to paint!). An art and a science. And it doesn’t happen as most would assume.
Rob
He is no fool who gives what he cannot keep to gain what he cannot lose.
5 Likes
You miss the essential fact that we make progress. Our ability to match color from run to run is probably like 5% of the problem that Ford etc. had to solve when building cars 100 years ago.
This is part of what makes manufacturing hard. Every step adopted to solve some problem at some cost has to be reexamined as the problem that step solved disappears. Using crowbars to align parts on the assembly line disappeared as we got better fit tolerances out of the basic manufacturing. Toyota improved its ability to get tolerances right in the first place years before the American manufacturers bothered doing so, and as a result they were able to DESIGN a better car because the tolerances they were limited to were better.
By taking a revolutionary top down approach to manufacturing, infomed by current capabilities, Musk has pushed aside piles of old techniques and piles of old limitations to design compared to the legacy automakers.
I remember the puzzle of Honda, which (sorta) pioneered transverse mounted engines in cars. Well they also pioneered in-line mounted engines on motorcycles. In some sense the real answer is you keep doing lots of things lots of different ways, and then you have the opportunity to keep the innovations that really matter which learn from your experience with those innovations.
The best way to build a car is a fast moving target.
6 Likes