Cory Steuben from Munro Live is interviewed. I thought the most interesting bit was where he talked about the GigaPress. For all its advantages, he explains that it is worthless if you can't get the throughput high enough. He says an auto plant typically puts out a car every 45 seconds.
This sounded low to me, but I assume he was taking multiple production lines into account?
So if the GigaPress can't keep up with the production rate then you might as well not use it. In this case, you need it to output a casting every 45 seconds or buy more GigaPresses.
As the resident Gigapress here on the forum I want to add to this.
Manufacturing lines in general need to have smooth flow through the production system with minimal inventory, which is achieved by matching the cycle time of every production step so that everything remains synchronized. In car manufacturing, around 45-60 seconds is the typical cycle time as Cory said.
However, steps can have a longer cycle time than the rest of the production system as long as they can be parallelized. Cory’s example numbers were:
“What it really comes down to is: can your throughput match your production? And so, getting the cycle time down in the area of a minute, now you’re matching the speed of the assembly line in final assembly. So essentially most plants will put a car out every 45 seconds to 60 seconds, and so if the machine is taking 120 seconds, right, you need twice as many machines and then it becomes less valuable. So almost more important than the capability of the machine is the ability to match throughput and that’s something that’s kind of overlooked. So the fact that a machine that large can output castings that fast is more impressive than just the scale of the machine, and the reason I mention that is if it took 5 minutes per part it would never make sense…”
In Cory’s example, if Tesla’s factory operates on a 60 second cycle but a Gigapress spits out a casting every 120 seconds, this can be balanced out by having two Gigapresses that
collectively output a casting every 60 seconds. In practice, incongruent cycle times aren’t always a clean whole number multiple like 120/60=2, but this can be overcome by rounding up and then having periodic machine downtime or slowing down their operational speed.
Analogy: A fast food sandwich shop will typically have one production line for assembling the sandwiches sequentially, but will have like 4 toasters. This is because the cycle time for each toaster is roughly 4 times longer than the cycle time for the sandwich assembly and cash register, so for smooth flow without a toaster bottleneck there must be multiple toasters that can work simultaneously in parallel and feed into the next step.
Ideally you’d need only one gigapress for the front casting and one for the rear and fortunately it appears IDRA and Tesla have achieved this. However, I’d bet the business case would still be strong even if Tesla had to buy twice or three times as many gigapresses per line because the savings are so large from eliminating 30% of the body shop, getting better first-pass quality and reducing vehicle weight.
Also, 45 second cycle time is actually pretty fast. 31,536,000 seconds in a year. With maybe 75% overall uptime including daily breaks, holidays, maintenance etc. , that’s half a million cars annually per line.