Designing the production system of a new product is at least an order of magnitude or two orders of magnitude harder than designing to initial prototype. In America, there’s less importance placed on manufacturing and I think this is a mistake. I sent an email to SpaceX saying this, I think designing a rocket is trivial. There’s tons of books. You read them. You understand equations. You design a rocket… The next step is you want to make one of those things. Making one of those and getting it to orbit is hard. The designing is not hard. The making of it is hard. The making of a production line that builds and launches many is extremely hard. And the next level beyond that would be creating a fully reuseable system, and have that in volume production and launch. That’s super, super hard. By building the machine that builds the machine, I mean the production system
Starlink production is going well. That was a hard thing to get right. We had many iterations on the starlink prototypes. Then building the production line was, I don’t know, a thousand percent harder than designing the satellite to begin with. It is important to design for manufacturing to have a tight feedback loop between the design of the object and the manufacturing system. When you design the object at first, you don’t realize all the parts that are really difficult to manufacture.
So have the design and manufacture, and bring those up at the same time so that in the beginning your going to make a thing that you know is wrong, but you’re figuring out what’s hard to manufacture. That’s the real problem. So we brought up the starlink production line before we had the design finalized, which is actually the right thing to do. We discovered there are all these things in the design that were difficult to make, therefore we must change the design. The satellite had the same capability, but it was just very easy to make and launch. The satellites are being produced at a rate faster than we can launch them. The cost of the satellite has dropped below the cost of transporting it to orbit… The cost of the satellite will keep coming down as we ramp up rate and make design improvements.
That was Elon Musk at the 2020 Air Warfare Symposium. Certainly, Engineering and Manufacturing Design is supposed to account for producibility in defense systems. However, the milestone acquisition process is linear, from prototyping to development to production. There is not continuous feedback from production units back into the design and development. That only occurs by accident, such as in F-35 LRIPs lasting well over decade. This “concurrency” between R&D and production was famously called acquisition malpractice by former then-AT&L Frank Kendall.
Yet as we’re starting to see in software, products should be in development and production simultaneously to iterate, learn, and improve. This seems to rekindle the 1950s obsession with concurrency over stage-gates. But it doesn’t mean production should ramp up at the start of prototyping or development. Perhaps what is needed instead is a subjective feeling of whether things are going OK, and launch production when development is only 50 or 80 percent complete.
Leave a Reply