Hacker News Reader

Comment by titanomachy | Hacker News Reader

37 minutes of rework per car doesn't sound like very much. It would be interesting to know what fraction of the total time that is.

View on HN

No, but if 85% of cars need it that adds up fast.

Is this expected to be how things are going to go for the foreseeable future? If the situation is improved quickly, I don't see how this matters at all.

They made a big deal about their push and how they made the number of cars they said they would (finally).

Now we find out that there was an asterisk to that. The cars weren’t ready for sale without lots of manual rework.

It makes their statement/accomplishment look much worse than they tried to play it.

“85% of cars need manual fixes” is not a good look for any automaker. But it’s especially bad for one that there are lots of questions about who tried to use that goal as proof they were last their issues.

Exactly.

Tesla says they pay $25/hour at Gigafactory, glassdoor says they’re paying $17-$22 hours to manufacturing workers.

I’m not into manufacturing, but extra $10-15, for a car with starting price $35k, looks normal.

It's also fixing a manual part, not an already automated part. Remember, Tesla has a much higher automation factor than everyone else, esp. to get the quality up and costs down in the end.

Fixing just this failing manual part is not the drama the competitors are declaring. Tesla also needed this artificial 5.000 number to reach their mandatory milestone.

As noted in other comments, other major carmakers can complete a car in about 1-3 minutes, not including customization tasks like painting.

So 37 minutes of rework per vehicle means that they're less than 5% as efficient as their competitors, despite the massive investments they've made their factory line. It's indicative of serious production issues and will be a major roadblock to scaling future production, since it means that if they don't solve the production issues leading to the rework, then they would need to spend 20x what their competitors to achieve the same volume of production.

EDIT: The 3 minutes is time-between cars. Each car takes 17-18 hours to get through the entire line. So 37 additional minutes is still pretty bad, but it's not 20x as inefficient on a time-basis. However, given the number of workers at the Tesla factory, they are still 20x as inefficient on a labor basis.

That's pipelined time, not per-car time. Just because a car is produced every 3 minutes does not mean it takes 3 minutes to produce - that would be incredible.

Are you sure? Not a production expert, but it sounds to me like this is a confusion of 'complete a car every 1-3 minutes' and 'complete a car in 1-3 minutes' They are two very different things, and I'm quite sure that no one is completing entire cars in 1-3 minutes.

It's not bad at all, because it is only needed once until the bug is fixed. It's only a wrong manual task, so fixing this task sounds even trivial.

The problem is that it's not one particular problem with the assembly of the car--it's many problems that all need manual fixing. ("Manual task" in Teslaspeak refers to any task performed by a human on the line, not to any particular task.)

37 minutes * 4300 cars = 2,651 hours of work.

So basically one full time employee for a year.

It depends on what parts need rework. It may not be doable by one employee.

If it's $80K a week, isn't that more like ~50 full-time employees for a year?

Oh I didn't realize it was per week. You're right.

2,651 hours/week * $30/hour = ~$80k/week. on a volume of cars that brings in $200m/week. so what? its a minor inefficiency.

not really. each rework is the equivalent of a branching prediction failure in a very very long cpu pipeline.

all tasks are massively parallelized in a production line. to reach 5000 car month tesla need to crank a car every seven minutes.

now imagine a production line that pump a car out every seven minutes and you need 37 minutes manual labor for each of them

the inefficence doesn't look so minor. it's also a logisticsl nightmare, just imagine parking all those cars while the rework stations free up.

In the time it takes one person to do the 37 minutes of rework you have taken 5 more cars off the line.

So 6 mechanics in 6 bays at the end of the line do the rework jobs and cars continue to flow out of the plant one every 7 minutes.

An extra $20 in COGS per car. Is this really so horrible? What am I missing?

Even if it’s 10 mechanics and 10 drivers moving cars into and out of the bays, with some bays idle some of the time and let’s call it an extra $60 in COGS.

Their gross margin is already way better than anyone expected to see on the Model 3, they used some very impressive electrical design to save in so many places.

Even just their rear view mirror is more than $50 cheaper than the competition’s mirrors because of their smart electrical design. [1]

Right now they need volume more than anything else. Speaking as a reservation holder.

[1] - https://www.teslarati.com/tesla-model-3-solidly-profitable-s...

> What am I missing?

just on top on my mind tooling, factory space, training, burnout, turns, above average defects that will have to be handled by the post sales network.

> they used some very impressive electrical design

oh yes they are brilliant car, no question about that. it's less impressive when you get an engineering marvel and the build quality is random depending on the hands that worked on it in post processing.

> the inefficence doesn't look so minor. it's also a logisticsl nightmare, just imagine parking all those cars while the rework stations free up.

To park a car every seven minutes? Maybe one extra full time employee?

> logisticsl

(and btw it's 3x that because full time still only means 8 hours a day)

Per employee doing rework. To complete tasks which involve taking pieces off and putting them back on in ~30 minutes, you're probably looking at up to 10 people (for example, changing a bushing on a Mercedes that requires the complete disassembly of the rear of the car is around 6 hours of labor).

As a point of comparison, large car manufacturers get into the "how much would a lawsuit and recall cost" when discussing adding costs of $4-$5 per car.

Per [0] it seems like line workers make ~$20/hr, so that's over $50,000 per week to fix the cars, not including the cost of parts or associated delays. It's only about $12 per car, but for Tesla I imagine every dollar counts.

[0]: https://www.indeed.com/cmp/Tesla/salaries?location=US%2FCA%2...

That's assuming the cost of labor is the only operating factor here. There are assumedly parts, factory operation, other opportunity costs at play here.

Also, since Tesla has all of these cars under warranty themselves, if they do have issues after delivery, which quality control problems like this usually indicate it can really blow through their margins quickly.

It's a bit more than the output from a single person in a year. You just have to hire 2 more guys and you're good.

It takes 9 months to make a baby. You can't get 9 women and produce a baby in a month.

2 new guys don't have a year to rework 4300 cars.

Reply on news.ycombinator.com