Elon Musk and the physics of Tesla’s Model 3 factory

The machine that makes the machine is Tesla’s challenge.

Just four days after Elon Musk delivered 30 production versions of the Model 3 to an adoring crowd on July 29 he reported Tesla's second-quarter results. The numbers weren't as bad as they could have been: The company merely lost $336 million, and the stock went up anyway.

But there is a curious phrase in the first sentence of Tesla's earnings note that has been not discussed much: "During Q2, our engineering, manufacturing and supply chain teams were focused on the final stages of Model 3 product development and building the 'machine-that-makes-the-machine' for the start of production."

The what? During an investors meeting on June 2, 2016, Musk talked about "my favorite thing: applying physics-first principles." In this case, he was applying them to the Tesla Factory. Musk told the assembled that he'd been spending a lot of time in the plant, not at a desk or in an office, but on the floor itself. He came to recognize the factory as "the machine that makes the machine."

He said that when Tesla designs a car, it doesn't go to a catalog. Rather, Tesla designs what it thinks is necessary, then develops or sources the parts. He suggested that the same thing needs to be done in terms of equipping a factory. Musk concluded that while vehicle engineers are chasing tenths of a percent improvement, in the factory, there is "potential for improvement by a factor of 10."
He expounded on that, noting that for a given size of a factory, there are three factors: volume, density and velocity. As for volume, there is but a small percentage of the total available used.

Presumably he was talking about all of the space that doesn't seem to be used above the floor, above the height of the highest machine. Ceilings tend to be high in factories for such things as overhead cranes and material handling equipment that carry things like painted closure panels. So it's not like factories are volumetric vacuums.

As for velocity, he said that the best factories in the world make a car every 25 seconds. Taking the size of a given vehicle and some buffer space, Musk reckoned that the required space is five meters, so if a car exits the line every 25 seconds, this means that the movement is on the order of 0.2 meters per second. He suggested that this is the rate that a turtle moves. So much for fast factories, right?

Density brought him to talking about the design of computer chips, where there is tremendous density, with things exceedingly close together and several lines and layers. He said that the density of the factory is on the order of two to three percent and that it was possible to increase that by an order of magnitude, to 20 to 30 percent.

At present, Tesla has net orders for approximately 455,000 Model 3s. In Q3 it expects to have a production rate of 1,500 cars per week, going to 5,000 per week by the end of 2017. Then, according to Musk, they will hit 10,000 per week "at some point in 2018."

Which brings us back to the Tesla Factory. Prior to Tesla's acquisition in 2010, it was the NUMMI plant, a joint venture between General Motors and Toyota. The factory had a standard annual capacity of 385,000 vehicles. They built Novas and Prizms, Vibes and Corollas, and even Tacomas. In 2006 they produced 428,633 vehicles (clearly there was a whole lot of overtime that year). Toyota essentially ran NUMMI ( GM personnel were there to learn). And when it comes to optimizing the "machine" that is the factory, there is no better company than Toyota. (It is, perhaps, no coincidence that the Bible of Lean Manufacturing, The Machine That Changed the World, is essentially about the manufacturing machine that is Toyota.)

There are two ways of looking at Musk's factory physics. If we look at standard 385,000-unit output of what had been the NUMMI plant and increase it by a factor of 10, then the Tesla Factory ought to be making more vehicles than 10,000 per week: that's only 520,000 units per year, when a factor of 10 would have it at 3,850,000 or 74,038 cars per week. If we take the full production to be 520,000 units, then given that that is only a 15-percent improvement over the 428,633 that NUMMI produced in 2006, this is not exactly a high-performance machine. Tesla is defying gravity with its valuation. But when it comes to actually producing cars at high volumes, it has yet to gain momentum. Or figure out the proper math.

Related Video:

Share This Photo X