Hacker News Reader

Hacker News

Comment by Aloha | Hacker News Reader

Full thread

Aloha

As someone who is familiar with doing things at volume - I dont see how this makes it actually cheaper -

* It doesn't reduce labor costs

* It doesn't reduce materials costs

* It doesn't seem to allow them to assemble faster

* It has significant challenges for quality (matching paint across body panels is very hard, for example)

The only thing it does seem to reduce is factory footprint, which doest seem to be a meaningful cost on a per unit basis (a single factory will build millions and millions of vehicles over the life of the factory).

So what does it do, and why is it better?

View on HN

The way I interpreted the short amount of text I could see, was that it would ideally enable better concurrent and parallel execution of part fabrication, as in initiating computation on multiple threads across multiple cores, and then sending the result to the main thread for UI (paint and layout).

I have zero knowledge of manufacturing, but if that worked out, would it not increase the efficiency of converting raw materials into a final product and yield cost savings by minimizing controllable downtime? I'd think you'd just hit external bottlenecks in the supply chains, but in my imagination that's where you'd want them to be.

Speculating but I think a production line has a rigid synchronized speed. If a step suddenly takes more time everything else must wait.

Imagine a queue of people at a buffet where overtaking is not allowed. If someone dawdles around at the juicer, people behind them need to wait while the salad bar is emptying and suddenly the cashier has nothing to do.

That adressing throughput not costs. If you keep costs contant, lower prices but acchive higher throughput, the only thing you acchieve is to lose money faster than you would have done otherwise by just lowering costs straight away.

Many years of Japanese factory management techniques have demonstrated quality and efficiency in the auto industry. They work closely with the line by making sure consistency is maintained to a high degree in the queue and by stopping the line when it needs to fix the problem to keep problems from compounding downstream.

Adding a capacity for parallel fix up product looks good temporarily but ends up decreasing quality and productivity with a lot of extra waste - one thing that Boeing has recently shown again.

If a station has a longer cycle time than others on a line, you just duplicate the station so that the throughput remains the same. If you have say 10 painting stations, you can on average pass a car through in 1/10th the time it takes to paint a car.

This makes sense when everything is running normally, but if one of your 10 painting stations goes down for a day, every upstream assembly step is forced to run at 90%, since the painting step is now a bottleneck. Given the huge cost of idle capital, it makes sense to minimize the upstream impact of any given issue.

And if any one of your single stations goes down the line is forced to run at 0% speed. You add sufficient redundancy and perform preventative maintenance so the line doesn't go down. 11 car painting stations is still way easier, cheaper, and safer than building 1 station which is 10 times faster that can never go down.

Paint operations are typically one of the major constraints on production flow. Paint facilities are expensive and require lots of environment review and hazard remediation. It’s probably cheaper to paint at the assessmbly level in China with no scrubbers or expensive facilities but getting matching color is hard

It raises the stock price.

Likely the most accurate reason.

Tesla is on-track to miss delivery targets which is why we've seen a lot more PR articles recently.

As well as initiatives like forcing everyone who test-drives or services their car to try FSD which will help justify poor numbers.

I think it does allow them to assemble faster. I found another article here: https://www.nextbigfuture.com/2023/09/tesla-will-build-a-car...

It claims that it reduces the number of parts (and presumably simplifies assembly)

Any sources for those claims?

Shouldn't it reduce labor costs when you have just one part instead of hundreds?

How does this reduce the number of parts on a car?

Like, okay, you currently build a vehicle by dragging a hulk down a moving line.

This says leave the hulk in one place, and bring the parts to it.

It on its face appears to be less efficient.

It gives the appearance of innovation and disruptive thinking to people who don’t know better.

It's not as if legacy automakers worry about production efficiency and quality...

Reply on news.ycombinator.com