Air cooled battery modules

[Genshi]

Having ruminated on it a bit, I find his arguments compelling, but I'm not won over. As many of you noted, the air cooling wouldn't dissipate heat as well (obviously) but it might dissipate it well enough. That said, since all 3 coolant systems in the S are tied together to some extent, It does seem like an odd departure. Not knowing enough about charging & heat generation/dissipation in batteries of this size and power, I honestly don't have the education or knowledge to make a coherent guess either way.

I do suspect he's right on one overarching notion though.... Tesla & Panasonic are substantially ahead of the competition where battery packs for BEV are concerned. Whether that means they figured out air cooling & active balancing paired with the confirmed 20700 cells, or they've found a cheaper way to implement liquid cooling, or it's something they've figured out with battery chemistry or something else entirely remains to be seen. But by all reports their costs pretty much have to be lower by an order of magnitude than anyone else's to be able to produce the Model ≡ at a healthy profit margin.

One thing that occurred to me though: Analysts & fans watch Tesla very, very closely, and that often includes any patents they're filing. It seems odd to me that no one has recently spotted anything related to new or innovative air cooling design. I know the liquid cooled designs all have patents associated with them, but I'm unclear on when in the process of development those were filed and too lazy to google-fu my way to an answer.

 

[stopcrazypp]

That said, since all 3 coolant systems in the S are tied together to some extent, It does seem like an odd departure. Not knowing enough about charging & heat generation/dissipation in batteries of this size and power, I honestly don't have the education or knowledge to make a coherent guess either way.

I thought of that first reading the title, but if you look at his proposal in detail, actually that part is irrelevant. The interface between the battery and car could be kept the same as in the S today, but instead of running the coolant through the cells, the coolant is only run through the central heat exchanger/radiator.

This is exactly the same way the Nissan e-NV200 active air cooled pack is designed.
Note the two silver coolant tubes near the middle that go to the radiator (on the right of the picture) inside the pack. Then a fan is used to circulate the air through the radiator.

Does Nissan e-NV200 Show Change of Policy on Battery Heating, Cooling?

 

[Cosmacelf]

Frank, the article author's main point wasn't so much about efficiencies from battery balancing, it was about how you could skip a very expensive cell manufacturing step if you included active battery balancing electronics into the pack.

 

[Genshi]

I thought of that first reading the title, but if you look at his proposal in detail, actually that part is irrelevant. The interface between the battery and car could be kept the same as in the S today, but instead of running the coolant through the cells, the coolant is only run through the central heat exchanger/radiator.

This is exactly the same way the Nissan e-NV200 active air cooled pack is designed.
Note the two silver coolant tubes near the middle that go to the radiator (on the right of the picture) inside the pack. Then a fan is used to circulate the air through the radiator.

Does Nissan e-NV200 Show Change of Policy on Battery Heating, Cooling?

That's funny, I'd just looked at that article as I was contemplating the merits of the OP's link. It did, from what I'd read/seen of the Model S seem like the coolant systems could be broken apart, but with such an elegant system already in place, it was hard to imagine. Then again, this is a company that works from first principles on pretty much everything, so it's not like anything they've created would be out of bounds for a redesign given proper justification.

 

[Frank99]

...how you could skip a very expensive cell manufacturing step if you included active battery balancing electronics into the pack.

No argument here - but he spends pages and pages describing matching, when it's fairly unimportant. Being able to skip the separate charge/discharge cycles necessary to stabilize the battery pack is important; matching the cells and balance charging (and hence the active balancing electronics), not so much.

Cell matching is only important for series-connected cells - that's where you run the risk of overcharging or overdischarging individual cells. The important thing to note is that Tesla's packs are groups of parallel-connected cells, with the groups connected in series into modules and modules connected in series into the overall pack. Thus, it's important that each group be relatively well matched to every other group - but the fact that the group is composed of 74 parallel connected cells means that you're essentially matching the average of a group of 74 cells to the average of the 74 cells in each of the other groups. This drastically reduces (if not eliminates) the need to do matching and active balancing.

Because of this averaging effect, I believe Tesla would be able to build the cells into modules or packs, and run the conditioning cycles on those without having the complications of the active balancing electronics. The simple group balancing that I discussed before is sufficient. This whole process wouldn't work if there were a measureable failure rate on individual cells - if you tried building modules from untested cells that average 98% "good", you're building in one or two cells to every group that will need to be manually replaced. That's not good.

 

[Yggdrasill]

I found it funny that at first he assumed $50 per cell for the production cycling hardware. Then later he said that the hardware for cycling 2880 cells in the pack would cost $660.

$50/cell or $0.2/cell - which is it?

 

[stopcrazypp]

I found it funny that at first he assumed $50 per cell for the production cycling hardware. Then later he said that the hardware for cycling 2880 cells in the pack would cost $660.

$50/cell or $0.2/cell - which is it?

$50 per cell is his estimate for the hardware in the factory as an example of the conventional process. His $660 estimate is for the dynamic balancing hardware in the battery pack as an example of his proposed solution.

Reading the comments now in the article, others pointed out the same thing I did. The biggest problem with his proposed solution is that it is unlikely that the per cell evaluation steps can be skipped simply for safety reasons (even if the pack fuse will disconnect a defective cell, an internal short can still cause a thermal runaway).

 

[Yggdrasill]

I decided to look at how a cell cycling facility for 11 million cells would look like, and it doesn't seem too infeasible or expensive.

At the end of the cell production line, you could stack the cells into a 1 meter by 1 meter by 7.5 cm tray. I think you could stack something like 2000 cells into such a tray without great difficulty. This tray could be carried to a stack of cell testers. Basically 1.1 meter by 1.1 meter by 30 cm modules, stacked 5 meters high, with a slot for the cell tray, and robust pin connects at the top and bottom. After the tray is correctly positioned in the cell test module, the interface is raised from the bottom and lowered from the top, connecting all the cells to the testing module. All 2000 cells are connected to a circuit board, which cycles the cells correctly, then assesses the quality of each cell.

Each 5 meter high and 1.1 meter by 1.1 meter stack would hold approximately 30.000 cells. And you'd need about 370 such stacks, or about 5550 cell test modules. Assuming each module costs 5.000 USD, and each stack adds 50.000 USD, the cost would add up to approximately 46 million USD. And the required space would be somewhere in the region of 600 square meters or 6500 square feet (assuming a fully automated facility), with 5 meter high ceilings.

 

[cronosx]

Just a note.. the design/space he has calculated is based on the base battery ( if i read right ), but sure as hell, there will be a battery with more capacity.. so the you'll need more space, since with the design he has in mind we have just used all the available space.. where do we put the extra cell?

 

[Jayc]

I'm just curious if either of you read the article?

Yes I did but what can happen sometimes is we get too engrossed in the technical side of things that we completely ignore the obvious.

1. R&D for Model 3 is mostly covered by way of the wealth of knowledge gained doing Model S
2. Model S will always be at the forefront of technology so any drastic tech change will first be in Model S Gen 2

And common sense will tell you there is a big risk involved in trying a new, unproven, battery cooling method on a car for which they can have almost half a million of orders within its first few years. The more obvious solution is to improve on what they already know.

In terms of improvements, I would look out for design aspects of Model S that didn't quite materialize as originally intended - Tesla will try to improve on them. Take for example Tesla's original goal of facilitating battery swap within minutes - yes they achieved that goal but in the end it became a non-existent use-case. So it is possible, this time they are creating a battery layout that is easy to troubleshoot rather than swap out. So for example, if the Model 3 battery could be "serviced" by replacing sections of it very easily, that would make more sense in a mass market car targeted for lower cost.

 

[Yggdrasill]

Just a note.. the design/space he has calculated is based on the base battery ( if i read right ), but sure as hell, there will be a battery with more capacity.. so the you'll need more space, since with the design he has in mind we have just used all the available space.. where do we put the extra cell?

He was assuming the largest battery would be around 65 kWh and offer a range of 320 miles. I think this is optimistic.

I think we can assume around 55 kWh for the base version offering around 220 miles range. For 320 miles range, you'd need something like 80 kWh. This means that instead of 2880 cells like he assumed, the battery pack may need to accomodate closer to 3500 cells. The energy density may also be optimistic, so something like 4000-4500 cells is also possible. This would mean that there wouldn't be as much spare space as he is assuming.

 

[Swampgator]

MODEL S 60 gets 210 miles with 60 kwh, and weighs around 4600 pounds.
A 55 kwh battery will not be needed to achieve 215 mile range in a 3500 lb car with a lower cd than Model S

 

[JeffK]

MODEL S 60 gets 210 miles with 60 kwh, and weighs around 4600 pounds.
A 55 kwh battery will not be needed to achieve 215 mile range in a 3500 lb car with a lower cd than Model S

Meh you still might be looking at over 250 watts per mile minimum and that's only if it's more efficient than a lighter weight BMW i3. Granted the BMW i3 has more drag.

While the cd will help for highway driving it's not really going to do a whole lot for city traffic where you're still accelerating a mass from a stand still. 55kWh would be a bare minimum unless something has dramatically changed.

 

[stopcrazypp]

MODEL S 60 gets 210 miles with 60 kwh, and weighs around 4600 pounds.
A 55 kwh battery will not be needed to achieve 215 mile range in a 3500 lb car with a lower cd than Model S

210mi/60kWh, suggests 192.5 miles of range for 55 kWh. A bump to ~220 miles would represent a 14% increase in efficiency.

I did the math here before comparing the car to the i3 and shows 55kWh is about right to get comfortably above 215 miles. The Model 3 will get better highway efficiency than the i3 to make up for the city efficiency (which is going to be practically impossible to beat).
Tesla confirms Model 3 will have less than 60kWh battery option

 

[McKricas]

I agree with almost everything in this article.

My two comments:

1. "Air Cooling" and "Heat" - With new 20700 batteries from Tesla/Panasonic for M3 one should be aware that an improved chemistry will for sure be used. And there's a great probability that these new batteries will have a higher capacity and a lower internal resistance that will significantly lower heat losses and will allow for charging/discharging faster (using higher amps).

An "Air cooling" system, if well designed as the author referred, using very dry air, a sealed pack and a powerful blower that can flow air at 30mph at least in a lower pressure environment, will be able to cool the cells with high efficiency probably equal or better than liquid cooler. The only thing needed is sufficient space or volume to compensate the lower thermal capacity of Air. But as the author found out, the M3 modules sizes are particularly bigger than they should be if using current Model S/X technology. So, a bigger volume module is already there.

The better part, is that these systems will also not be hard to test and build - similar systems are used in lots of other places. This actually will be much simpler (and cheaper) than the current liquid-cooler Tesla is using now.

So, if we believe Tesla will have a better chemistry with better capacity and lower IR in their 20700 cells and they can design a good enough and cheaper air cooling system that will have an equal or better efficiency than current liquid cooling, I'm sure this will be more than just a mere possibility.


2. Dynamic (active) Balancing System - This is a very interesting thing and could be the "real" reason for Tesla actually building the "Gigafactory". Just think a bit about it. Tesla always said little or nothing about WHY the Gigafactory is such a better strategy. It has always referred generalities like better "economies of scale" by achieving a 30% reduction costs on "transportation" / "insurances" or by being able to make cell production lines bigger, etc. Nothing real technical or dramatically innovative.

An Active Balancing system actually makes total sense. It's true that if one looks at the way Li-ion cells are currently built, the cell forming/aging is one time consuming and reasonable costly step that if substituted or skipped would make overall cell production much cheaper and better.

If Tesla is really concentrating and focusing at "Module level" rather than at "Cell level" (which makes total sense) then the Gigafactory is suddenly explained and perfectly justified. This will be a kind of "killer advantage" that no other carmaker will have.

"Module level" implies that all cells are "dumb" and all the intelligence is in the Module control. Power is controlled and distributed in a intelligent way according with each cell real capacity. Charging, discharging rates and also Tests are continually controlled at Module level. Also, if anything goes wrong the Module control system will detect it and will take some kind of action.

This is what an advanced "Dynamic Balancing System" just is. Removing cell forming/aging step from production and improving the Module control Intelligence can only be done if the Cells are not the "main focus".

 

[gregd]

Isn't there still a fundamental problem with air cooling of cells (i.e. without an active heat pump of some kind), where one is limited to cooling no colder than ambient? Around here, it gets 100+ F, and if sitting in traffic on a black asphalt roadway, probably in excess of 140F. You can't get better (colder) than that by blowing more air.

As nice as it is to drive my Roadster with the top off, I, like my battery, much prefer an actively air conditioned environment over what passes for ambient air, even at freeway speeds.

 

[Cosmacelf]

Yes, that's why the author proposed not using ambient air, but a sealed chamber with cool air conditioned air.

 

[diamond.g]

So reverse the pump for winter to heat the pack?

No one has talked about supercharging yet...

 

[gregd]

Yes, that's why the author proposed not using ambient air, but a sealed chamber with cool air conditioned air.

Hmmpf, seems mighty inefficient, but I haven't read the whole article (refuse to be required to create an account for such things, when 75% of the screen is already ads and click-bait). Datacenters are moving to liquid cooling in order to not waste the energy moving so much air around. Why should cooling a battery be better the other way?

 

[stopcrazypp]

Isn't there still a fundamental problem with air cooling of cells (i.e. without an active heat pump of some kind), where one is limited to cooling no colder than ambient? Around here, it gets 100+ F, and if sitting in traffic on a black asphalt roadway, probably in excess of 140F. You can't get better (colder) than that by blowing more air.

As nice as it is to drive my Roadster with the top off, I, like my battery, much prefer an actively air conditioned environment over what passes for ambient air, even at freeway speeds.

Yes, that's why the author proposed not using ambient air, but a sealed chamber with cool air conditioned air.

So reverse the pump for winter to heat the pack?

No one has talked about supercharging yet...

Read my earlier post about how the e-NV200 cooling system works. The proposal by the article is essentially the same thing (except cell spacing is much larger and layout is different).
Air cooled battery modules

Basically the connections into the pack is exactly the same with two coolant hoses (one in and one out). The differences are:
- In Model S/X, the coolant is directly run through tubes that touch the cells directly, thus allowing liquid cooling/heating
- In the article, the coolant is run through a radiator in the middle and then fans circulate air through the radiator to cool/heat the cells.

The coolant would already be cooled/heated by the AC/heat recycling system, just like it is with the Model S. So the temperature has nothing to do with ambient air.

What is changed is that instead of using liquid, air is used to interface with the cells. Air has roughly 4x less specific heat than water, so there must be roughly 4x the volume of air moved to provide the same cooling effect. That is the reason the author suggests for larger gaps between the cells.