unfortunately Tesla has a bad enough reputation in the industry that there aren’t many people currently employed who would want to work there. they gobbled up a good number of people back in '07-'10 when the other automakers were shedding jobs, and burned a lot of those people out.
Yeah, pretty much. We all do pretty much the same, but man-hours can be different, depending on how you account for them. As a generalization, final assembly operations count on a painted body and closures, and all of the other components (tires, seats, instrument clusters, etc.) being there, at the line, ready to install. A particular product – regardless of complexity – has a given cycle time, say, 30 seconds. Each station then has 25 seconds to perform the job, plus 5 seconds of takt time. Anything that happens not at line side isn’t counted in the man-hours, because those are purchased in assemblies. Now, you’re talking cost rather than man-hours. No one cares about the man-hours required by Johnson Controls to produce a seat; only that it costs $xxx (their man-hours are their problem, fundamentally, although cost operations will try to negotiate cost based on that).
The “man-hours” that produce a vehicle neglect this. It’s kind of an “I, Pencil” situation.
Nope. As above, if you have 25 seconds to perform a job, then that’s what you have. The neat thing is, the scheduling system has the component that you need line-side already, for that particular car. If you install a manual seat or an electric seat, the time is the same, because any difference is already account for in the price of the seat. In the case of something like a sunroof, there may be an idle station for non-sunroof, but this is generally frowned upon. We can pump out a completely loaded F-150 in the exact same amount of time as an austere model.
That is interesting. It seems pretty wasteful to me, but that’s coming from a chipmaker background. A computer chip is a factory of sorts; bits go in one and and come out the other, with many, many processing stages in between. If you have a particular stage that’s only used half the time, you try to arrange things so that you only need half the units, buffering the product so as not to back up the pipeline.
In some idealized factory, where half the cars have sunroofs, I’d make the sunroof station half speed (with some cost savings) and alternate the cars. Cars still move with an average throughput of 30 s/station, but in this case it’s 60 s for two cars.
Well, I realize there are all kinds of practical considerations that prevent this kind of thing. Maybe you could pull it off if you had one huge factory with multiple parallel lines all making the same car but with different options. Though you would probably run into the same issue that chips do: simply moving the product around can cost as much as just duplicating the functional units. It may be that you could have just 4 sunroof stations in a line that did 8 cars at once (in some gigantic factory), but that buffering and moving the cars around is more expensive than just having extra stations that are idle some of the time.
that’s not how assembly lines work. a “half speed” sunroof station would either slow down the entire line (increasing costs) or require diversion of entire vehicles off of the main line, then re-integrating them later. Yeah, try managing that.
it’s simply easier and cheaper to have a vehicle just go past the station if it isn’t getting a particular feature. and it’s not like there are a whole lot of cases of that where it would justify altering the layout of the line or plant. as Balthisar said, a lot of the differences between low-spec and high-trim vehicles are in the sub-assemblies. A bare-bones F-150 XL gets a dash, floor covering, seats, etc. just like an F-150 Platinum. You already know when each truck is going to arrive at a particular station on the line, so you sequence your parts delivery to the line such that when the XL arrives, the dash ready to install is the one with base radio, base cluster, and plain black trim, and when the next truck arrives (the Platinum) the parts carrier has the dash with the leather covering, chrome trim, LCD cluster, and SYNC 3 ready to be installed.
that would be a complete waste of space, time, and money.
edit: the only case I’m aware of where major sections are diverted from the main line is doors. The plants I’ve been in, once the body comes out of the paint shop, the doors are removed and hung on conveyors on the door/trim sub-line. then all of the stuff gets installed in them like the glass, window regulator, speakers, wiring, trim, etc. then they merge back into the trim & chassis line and are re-installed back onto the vehicle.
At what stage of the assembly line process is a VIN# assigned?
the VIN is created when the vehicle is scheduled for production. could be weeks in advance.
A major automotive innovation (started in Japan?) was bundled options. instead of a car with 20,000 option possibilities the consumer was limited to a much lower number. It lowered production costs and provide the buyer with more options for less money. things like remote adjusted mirrors became standard.
Without ever seeing a modern production facility I would imagine they cue up production runs in batches based on the limited numbers of combinations possible. Sunroofs will be part of a bundle that runs for days or weeks at a time.
I would imagine your choices dictate a variance of weeks. You might buy a red Model 3 when the current production is blue, setting you back 3 weeks.
I can’t see you buying a Bolt if you’re excited over a Model 3. The excitement isn’t going to be there for you.
nope. typically 4 weeks of production are scheduled at any given time based on dealer and customer orders. Which means that they know exactly what vehicle will be built when and with what options for those 4 weeks. so, if today, they’re only building 50 vehicles with sunroofs, then they’ll have only ordered 50 from the supplier. if tomorrow they have 100 vehicles scheduled which need sunroofs, that’s how many they’ve ordered from the supplier for tomorrow. the move to “just in time” logistics means this stuff is carefully planned out to the day. or even the shift.
as an aside, if you’ve ever wondered why it takes 6-8 weeks to special order a vehicle, it’s because you have to wait until they can schedule your order in the system after the current 4 week rotation is built. it only takes a day or two to actually build the car or truck; all of that time is simply waiting your turn in line.
aren’t you just restating what I said? production runs are based on the combinations ordered.
I don’t think I am. it sounds like you’re saying the content of the vehicle is what determines when its scheduled to be built; i.e. “we’ve got orders for 500 F-150s with sunroofs, so we’ll build all of those on Thursday and Friday.” If this is not what you meant please correct me.
production is continuous; there generally are no “runs.” and vehicle production is scheduled roughly as each order is received. To the best of my knowledge, the options the vehicle is ordered with has nothing to do with when it’s scheduled.
I’m suggesting, with nothing to back it up, that groups of options are run at the same time. That cues up the right parts at hand at all stations and a uniform installation.
But I’m not in the business. The purpose of bundled options was to increase efficiency in the production.
not in that sense. reducing orderable combinations reduces complexity which lowers logistics costs, and development and tooling costs can be spread out over greater numbers of parts. that and JIT logistics gets away from the need to warehouse tons of parts at the assembly plant. for something like the sunroof example we’re beating to death, that’s a component which a vehicle either gets or it doesn’t get. but the stations where the sunroofs are installed are always on the assembly line, if a vehicle doesn’t get one it just rolls on by those stations. there’s really no reason to group sunroof and non-sunroof vehicles; if your logistics and material handling people do their jobs right there’s always parts available.
for parts which are specific to a particular vehicle (e.g. a dashboard which has to have the right color, trim, and options for its target VIN) you just “sequence” them in the carrier when they’re brought to the line.
The nerd in my would love to see a modern assembly plant in action.
What jz is talking about for cars matches everything I’ve seen in visiting probably about a dozen or so final assembly plants for various types of aircraft. Except when the fundamentals of production were really significantly different - like Boeing’s 737 line is totally separate from their military P-8 line - I have not seen anything to indicate that manufacturers do things like “we build all the Black Hawk helicopters with sunroofs on odd days, then build the ones without on even days.”
Again, I’m just talking aircraft, but from what I have seen, the total focus is in giving workers the parts they need for that item at the right time to produce the right product; and not have steam whistles go off at various times where production of Variant A stops and Variant B begins, only to go back to Variant A in a little while.
Yeah–I wonder if the issue here is that vehicles are just so large and unwieldy. Manufacturing of small items deals with this situation all the time; if you have a station that is slow, for whatever reason, you just split the flow, parallelize the work, and reintegrate it later. Works fine for batteries or cookies or whatever; probably less so for cars.
The parts delivery stuff is interesting. This is one area where auto manufacturing *is *a lot like a computer chip. Many units are expected to deliver their results at a specific time and in a specific order. There’s no extra work in integrating the calculation results because it showed up exactly when you needed it.
It would be more efficient only if the costs of moving product around were low compared to the costs (human and machine) of a station. If land and building square footage were nearly free, that might be true–buffer zones, diverters, mergers, multiplexers, etc. would all be relatively cheap compared to how they are today.
Thanks for the insight, guys. Interesting to see how differing constraints (like the relative cost of transport vs. assembly) dictates the style of manufacturing.
Read this on the origin of Tesla
One thing that I found interesting from this story is that one of the hardest thing to get right in a car is the doors.
There is a lot more in the article but it shows the troubles with designing a radical new car.
The Germans are really, REALLY bad about NOT doing this. Option lists for BMW’s, Porsches, etc are eye wateringly long and can add considerable expense to a car.
Heck, you could start off with a 500 series Beemer that’s base price was 60k and easily get to 100k if you optioned it out. A lot of the automotive journalists regularly criticize them for doing this, with the logic being that a car that expensive to begin with should already include many of the options.
In most ways, what we’re describing is parallelization of the work. People tend to think of “building cars” as being that final assembly line, whereas, the final assembly line is simply the final integration of this huge system of parallel builds.
Think of the “I, Pencil” story for every part on the car. It’s easy to laugh at FCA for their quality, but it’s a huge feat of human engineering nonetheless.
“As Eberhard’s young company grew, he’d continue to ask would-be recruits to touch the dashboard, before throwing them into their seats with the torque of an electric sports car, properly unleashed.”
Carroll Shelby used to do the same thing to people in the first pre-production Dodge Vipers.
amazing how many things people think are new which have been done many times before.