REN: Tera factories

2021.06.30 – Ian Page

Musk pointed out that the world would need tens of terawatt hours of battery production by 2030.

That would involve hundreds of gigafactories.

So, he envisioned the tera factory - a factory the same size as a gigabattery factory that produced 30 times as many batteries from the same production lines.

He seems to be close to getting the production line running at the appropriate speed with a number of innovations in the 4680 cell. 

We have written about the materials supply chain issues and that most of the problems can be resolved with the LiFP chemistry, but there is still a problem matching the mining industry's speed of change with the battery demand curve.

However, in this note I will concentrate on what happens when a cell has been made.

If most of us think about it, we would expect some robot to whip the cells off the end of the production line and stuff them into a battery compartment which is then whisked away to be the basis of a car. Obviously, all the stages must be synchronized so that a big pile doesn't end up somewhere- the batteries are coming off the production line faster than a machine gun fires bullets, so any mismatch of speeds is going to be nasty.

However, the main problem is a stage called formation, which consumes about 30% of the capital in the factory!

The batteries are assembled into trays, and then charged and discharged a number of times. During this the internal structures necessary for battery longevity and capacity are formed. In addition, the behavior of the cells can be monitored to discover which are bad, and presumably identify best and worst of acceptable batteries.

There are millions of these cells, so it all must be automated to a very high degree. It also takes time to charge and discharge the cells.

The process involves producing heat during charging and discharging which must be removed - if local hotspots develop it affects the battery formation and could cause a disaster plant fire.

It also uses a LOT of electricity- its equivalent to charging hundreds of cars. Being smart the discharge of one batch is used to charge another.

With the cell trays packed to the ceiling and absolute minimum space between the rows, there is still a minimum volume required for a given number of cells and if there are 30X as many it seems that it’s going to be hard to avoid the majority of the factory space, most of the capital, and most of the work in progress being dedicated to formation. One could possibly move to a library stacks model where the shelves move and there is no space between them, but the wiring and cooling would need serious thought! This might save about 30% of the space.

Tesla has issued patents relating to the process, but I haven't seen any stunning ideas that would result in a 30-fold decrease in required volume per cell - there are some savings in corridor size and cooling systems and some simplification of the connections needed for charging and discharging but while being good approaches to improve consistency and reduce cost they are incremental not transformative.

As far as I can see the only potentially transformative solution is to dramatically reduce the cell charge discharge cycle time and the number of cycles needed.

Here we go beyond available information. So, treat only as a frame for estimating.

If 0-100% charging/discharging during formation takes 1 hour, and you need 5 cycles that's 10 hours.

To get the formation process to line up with a 30X production system, this would have to be reduced to 20 minutes or 4 minutes a cycle. For guidance the best estimate for 50% filling is 7 minutes and it slows down after 80%

Driving a battery fast involves wasted energy and that appears as heat, which would imply water cooling (which tesla has in a patent).

It's also not clear whether the necessary physical and chemical changes can be made to occur fast.

It’s conceivable that the formation process could in some way be eliminated - perhaps the cathode particles could be pre covered with the sei layer before assembly - but this is way beyond anything I have read about.

So, my current view (until Tesla release their brilliant solution) is that a terafactory is going to be a massive totally automated multi story formation plant, with lots of heat blasting off it, and a couple of small associated sheds where the batteries are assembled on a production line that looks like a coca cola plant, and the cars are molded in another.

It’s going to look very different from a traditional car factory!