Tesla CTO JB Straubel on Autonomous Vehicle and Battery Tech

[techmaven]

Lots of great info from JB Staubel on Tesla's perspectives on transport.

EV Annex blog post:

Tesla Co-Founder JB Straubel hints at forthcoming autonomous driving and battery breakthroughs [Video]

Video on Youtube:

Some interesting takeaways, in addition to stuff we mostly already know...

Things Tesla examined and likely won't come about that we thought might:
1) Vehicle to grid - seems unlikely from their perspective
2) Re-use of vehicle batteries in stationary storage - also seems unlikely

Both of these have a myriad of technical and economic issues that make them unlikely. Which then we have to look at Nissan and say, hmmm... someone is right, someone is wrong. I would hazard that Tesla is right and Nissan is wrong.

State of Model 3
1) They have talked to key suppliers already
2) From those talks, they believe they can double production from original targets
3) As a result, recent press reports about "supplier doubts" are smoke, not Tesla's suppliers which we suspected

State of Regulations and Standards
1) Needs work
2) Needs work
3) Charging standards need to accommodate future needs (duh)

Obviously, we already know the major points... electric mobility is coming due to technology changes, climate is driving the overall issue, and automated driving is also coming and is coming faster.

 

[JRP3]

Re-use of vehicle batteries in stationary storage - also seems unlikely

wk and others who are already doing so would disagree

 

[Krugerrand]

wk and others who are already doing so would disagree

Economically?

Yes, he said from an economical and technical aspect. For some individuals - yes. For the world at large - no.

 

[JRP3]

Well if you're going to be nit picky

 

[techmaven]

Yeah, from salvage vehicles with a great amount of effort, yes. Therefore it is really in the hobbyist realm. Ideally, we get less and less accidents where the car needs to be completely torn apart and therefore the economics of doing it en masse is unlikely fruitful.

 

[Cosmacelf]

From what I remember, the gigafactory is (will be) designed to recycle old battery packs. So they will have a semi automated battery pack recycling operation that will recycle the constituent materials into new batteries. I'm thinking that they would go all the way down to the individual metals in the cells. Ie. Crush/break open cells and go through a big mechanical and chemical separation process to extract constituent metals.

If Tesla does this, it will be interesting to see it. Normally that level of metal recycling is done in China with very little in the way of air quality and environmental safety. Would be nice for a US factory to do it properly.

 

[Krugerrand]

Well if you're going to be nit picky

LOL!

He did go in to detail about it and mentioned things like stationary storage needing higher cycle life than car batteries, uneven degradation of car batteries and wanting more predictability and consistency for moving them into a new/revamped use, and also that they expect 10+ years of car battery life, which then puts those 10 year old batteries behind the technology/chemistry/etc... curve for use as stationary storage. In the end it makes more sense on a broader scale to simply go new batteries for stationary storage and recycle the old car batteries - in case people don't want to watch the whole video that's a compressed version of what he said on the topic.

 

[JRP3]

From today's perspective it seems hard to imagine a pack with 70-80% of usable capacity being sent in for recycling instead of reuse, but since we are expecting substantial cost declines, density improvements, and cycle life improvements, messing around with 15+ year old packs, (13k mi/year x 15 year = 195k miles / 200 miles per charge cycle = 975 cycles), that aren't really meant for daily deep cycling probably won't make sense.

 

[Krugerrand]

From today's perspective it seems hard to imagine a pack with 70-80% of usable capacity being sent in for recycling instead of reuse, but since we are expecting substantial cost declines, density improvements, and cycle life improvements, messing around with 15+ year old packs, (13k mi/year x 15 year = 195k miles / 200 miles per charge cycle = 975 cycles), that aren't really meant for daily deep cycling probably won't make sense.

Yes, at first thought it seems a waste of that 70-80% and to see it go to recycling. At some point I'd expect we'll reach the limit of technology with this kind of battery such that it might then make sense to reuse them rather than recycle - but maybe then we'll be onto something else even better? J.B. gave the impression that they've revisited this topic a few times already and keep coming to the same conclusion. Perhaps they'll continue to revisit it periodically and at some point the conclusion will change based on where they're at in terms of technology and then it'll make sense? It's interesting to me that progress seems to be happening faster and to a great degree than I'd been expecting from what they've always said about battery/chemistry/etc... advancement. Or maybe time is just passing me by faster than I realize.

 

[Cosmacelf]

Recycling versus reuse. Think about your PC. By the time it is five years old, you honestly can't think of a good use for the now boat anchor. Too slow, clunky, ram and hard disk are too small, video card is ancient, etc. The PC still works. Hasn't degraded one bit from the day you bought it, but the vast majority of people end up throwing them out, or if they put in the extra time, take it to an electronics recycling place.

Same thing will happen with batteries, and kudos to Tesla for figuring this out. Maybe in the future when there are no battery advances for 10 years, things will be different, but for now, recycling down to the metals is the way to go.

 

[Fallenone]

I'm a bit surprise about the reuse of vehicle batteries. From a dissemble point, if they are going to recycle them, they need to tore the car's battery pack down to cells anyway. So this should not be a limiting factor. I suspect the car's battery pack is not for daily cycle use thus not good enough for the Powerwall after 8 years or so but not Powerpack?

 

[JRP3]

You can use faster, more crude methods if tearing down for recycling than re-purposing.

 

[Cosmacelf]

Well, I tried to listen to it. Got halfway and had to give up - generalities, apple pie, platitudes, etc. I miss the old JB that said technical things and the old Tesla that had technical blog posts.

However I did look for JB's answer to vehicle to grid and reusing batteries. He made a lot of sense. He said that batteries for grids needed to have 4-5 times longer cycle life than vehicle batteries. People generally don't realize that vehicle batteries don't cycle (ie. one complete 100% charge to 0% discharge, or even 90 to 10) very often. However grid batteries do, meaning that battery chemistry MUST be different for efficiency and economic reasons.

Vehicle to grid doesn't work for this reason (you are using a battery meant for lower cycle life and now are cycling it much more frequently) and ALSO the complexity of installation (eg. cut offs if the grid goes down, etc.).

And reuse, he implied that after 10-15 years, a vehicle battery's cells are going to be all over the place in terms of useful life left, so how do you make a product with that? In addition, electric efficiency (power in versus power out) goes way down on an older battery making it even less useful.

 

[neroden]

2) Re-use of vehicle batteries in stationary storage - also seems unlikely

Bluntly, I think this one will happen, but that it's too far away in the future to be relevant. The "problem", if you can call it a problem, is that the batteries have a very long life in the vehicles. They are going to be used in vehicles for *10-20 years*. There won't be a significant supply of used batteries from vehicles until the mid-2030s! This makes them irrelevant in the storage market until then.

And reuse, he implied that after 10-15 years, a vehicle battery's cells are going to be all over the place in terms of useful life left, so how do you make a product with that?

For the emergency-battery market (which will still be quite lively), as opposed to the overnight storage or grid-balancing markets (which are quite different), they don't need much useful life, they just need to work on demand once or twice. Test the cells and use the ones which are "good enough". The cells won't be worth much, but they only have to sell for more than their value as scrap material, which they probably can.

In addition, electric efficiency (power in versus power out) goes way down on an older battery making it even less useful.

Again, I think there's a market there, once there are hundreds of millions of used batteries coming out of scrapped cars every year. But there's no point until there's enough of a volume of batteries flowing through to set up a full-time battery "reuse factory" where the cells are tested and sorted. This is so far in the future it's not worth thinking about this decade.

PCs aren't really the right comparison, and the reason has to do with Tesla's commodity design of the battery cells. The packs will be worthless but it'll be tempting to reuse the cylindrical cells rather than remanufacturing them, and I figure when enough of the cells with completely identical form factors are coming in for recycling (and it has to be hundreds of millions) they'll do it. In fact old hard drives do get reused if they have absolutely standard form factors and interfaces, generally in low-usage-per-day, high-quantity-of-data, low-budget situations. And those of us with ATX form factor cases don't replace the cases very often. Standardization of parts and interfaces drives reuse.

 

[stopcrazypp]

From today's perspective it seems hard to imagine a pack with 70-80% of usable capacity being sent in for recycling instead of reuse, but since we are expecting substantial cost declines, density improvements, and cycle life improvements, messing around with 15+ year old packs, (13k mi/year x 15 year = 195k miles / 200 miles per charge cycle = 975 cycles), that aren't really meant for daily deep cycling probably won't make sense.

Yeah, in the context of the majority of the battery pack population (which is probably going to be 10+ years old at minimum by the time they are retired), those cells would be drastically outdated.

Right now a limited amount of cells/modules pulled from salvage vehicles is viable in a small scale, but not really in the large scale JB is probably thinking.

 

[JRP3]

Test the cells and use the ones which are "good enough".

But there's the problem, the time and effort involved in testing individual cells from a 15-20 year old pack will not be cost effective. Since individual cells can't be easily removed from modules without destroying the structure there is no point in even trying at the cell level, and probably won't be worth even opening up a pack and removing modules. About the only thing that might make sense it to take an entire pack that still has reasonable capacity and plug it in as a unit for occasional backup duties. Maybe.

 

[techmaven]

There are definitely people tearing down salvage Model S packs today and making a meager profit doing it. But the cells from salvage packs are usually pretty new. I think incorporating entire modules is definitely the way to go for cost effectiveness in the U.S. Outside the U.S., if the modules can be sent somewhere that the labor is exceedingly cheap, then per cell teardown might be worth it for relatively new salvage packs.

In the case of an old pack, the degradation curve is pretty flat and then nosedives. As pack degrades past 80% capacity, into the 70%'s, it isn't clear that there is a return on investment, no matter how low the acquisition and reclamation process might be. At that point, recycling is the best option. Even if recycling itself isn't economically viable, building in the recycling process means that the entire lifecycle can be as sustainable as possible where the total mineral input is as low as possible.

 

[JRP3]

In the case of an old pack, the degradation curve is pretty flat and then nosedives.

That seems to be in contrast to the cycling graphs I've seen for NCA cells, which basically goes flat after around 80%. Page 20 for example:
http://www.embedded-world.eu/fileadmin/user_upload/pdf/batterie2011/Sonnemann_Panasonic.pdf

 

[neroden]

But there's the problem, the time and effort involved in testing individual cells from a 15-20 year old pack will not be cost effective. Since individual cells can't be easily removed from modules

They can. In fact, they *have* to be for recycling... melting the whole module at once is not going to give good results. Mixing dissimilar materials is what you try to avoid doing when recycling as it's a waste of energy. A productive automated recycling method will be stripping the top of the case and then pulling all the cells out of it, dumping the case and the cells into separate streams.

If there are millions of identically-shaped cells coming through a recycling production line every day, the effort of testing them with automated equipment becomes trivial. There is literally no point in building such a thing with less than a flow of millions of identically-shaped cells, however.

 

[ggr]

They can. In fact, they *have* to be for recycling... melting the whole module at once is not going to give good results. Mixing dissimilar materials is what you try to avoid doing when recycling as it's a waste of energy. A productive automated recycling method will be stripping the top of the case and then pulling all the cells out of it, dumping the case and the cells into separate streams.

If there are millions of identically-shaped cells coming through a recycling production line every day, the effort of testing them with automated equipment becomes trivial. There is literally no point in building such a thing with less than a flow of millions of identically-shaped cells, however.

I agree that the testing could probably be automated. However, the individual cells may or may not be charged before they go into the tester. If they're fully charged, and measure a high enough voltage, great. But if the voltage is low, I think you have to try charging it before you actually know whether it's good or not, and that takes time. Whereas, putting the cell into a machine that pulls off the outer cover and separates the materials, or whatever, won't take very long per cell. I think it's better to just unconditionally recycle the cells.