Air cooled battery modules

[stopcrazypp]

The problem is temperature differential per volume per second. Since, you are not an expert on it, you can either take my word for it, or become an expert. You might get that kind of cooling by bubbling the air through liquid nitrogen, but otherwise, I don't see how to do it.

As to the small fan, those fans you found assume no back pressure, not trying to run sealed air volumes past vast areas of heat exchangers. Nor was I talking about physical size.

Well, I'm not the type of person that trust appeals to authority with no math, so I choose to trust the author who said he did the math rather than the two choices you offered (which is a false dilemma).

And again, you did not address my point about why the radiator in the front of the car can remove the same amount of heat through air (actually more, since it actually has to handle drivetrain/inverter heat) and why suddenly this becomes impossible as soon as it moves inside the pack.

There are equations available that can simulate the author's proposal (which involves blowing air around a cylinder), but I'm a bit lazy to find all the variables and do the math.
Churchill–Bernstein equation - Wikipedia, the free encyclopedia

To be clear, I'm not in favor of an air cooled solution (I prefer liquid and I wrote a long post about the major issues with the author's article), but so far I have not seen any definitive proof that the author's solution is impossible.

 

[Jayc]

This is the crux of the problem The contact between the component that is chilled and the cells are the key. More the contact better the cooling. Air is the best in that regard and not snaking pipes.

To add to what @JeffK has said...

A variation of the same problem exists in ICE cooling and as we all know, only a very few engines these days are air cooled so that would seem to indicate that water cooling works better. As has been pointed out, one reason could be that it is relatively easier to push a liquid coolant through engine components than air. Also don't forget that after heat is captured, the job does not end there, it has to be cooled down so would go through a radiator where water will be better. And I guess the other advantage is that this coolant medium remains a close circuit so will be immune to outside dirt and contaminants. I hope no one is proposing we pump outside air with pollutants, dirt, moisture and dust through the battery system.

 

[JeffK]

And again, you did not address my point about why the radiator in the front of the car can remove the same amount of heat through air (actually more, since it actually has to handle drivetrain/inverter heat) and why suddenly this becomes impossible as soon as it moves inside the pack.

A radiator in the front of a typical car is a prime example of liquid cooling. Liquid moves heat away from the engine as fast as possible, transfers heat to the radiator via conduction, and air passing over the fins of the radiator allow heat transfer to the air over an extremely large surface area via convection and radiation. Liquid cooling in a PC works the same way...

 

[Topher]

And again, you did not address my point about why the radiator in the front of the car can remove the same amount of heat through air (actually more, since it actually has to handle drivetrain/inverter heat) and why suddenly this becomes impossible as soon as it moves inside the pack.

Yes, I did, you just didn't understand it. The problem is temperature differential per volume per second. The radiator has an infinite supply of air at ambient temperature. The proposal is for an extremely limited amount of air being recirculated. You can always increase the flow over a radiator, if it starts to overheat, cooling is less per cubic foot of air, but more cubic feet overcomes this. With a fixed amount of air this doesn't work, since if you don't cool it down, it comes back hotter 4 seconds later.

I'm not the type of person that trust appeals to authority with no math, so I choose to trust the author who said he did the math

So you'd believe me if I told I did the math (on the heat exchanger, I already showed you more math on the air flow temperature differential)? Didn't think so. Did he show you his math for the heat exchanger? No of course not. If he is capable of designing high performance heat exchanger why is he working as a financial journalist? He didn't even do the math I SHOWED you.

Thank you kindly.

 

[stopcrazypp]

A radiator in the front of a typical car is a prime example of liquid cooling. Liquid moves heat away from the engine as fast as possible, transfers heat to the radiator via conduction, and air passing over the fins of the radiator allow heat transfer to the air over an extremely large surface area via convection and radiation. Liquid cooling in a PC works the same way...

I understand all that. But that has nothing to do with my point. My point is concerned about the amount of heat being able to removed from the inside of the pack using air.
Here are the interfaces:
1) Heat transfer from cell to air
2) Heat transfer from air to radiator inside pack
3) Heat transfer from radiator to coolant inside pack
4) Heat transfer from coolant to radiator in front of car (plus perhaps an AC/heating in between)
5) Heat transfer from radiator in front to air

#3-5 there is no dispute is possible, since that pretty much exactly like the S/X pack today. #2 I don't see how it is impossible, since it is basically the analog of #5.

#1 so far no math has been done on this, but I believe the Churchill–Bernstein equation I linked would give a decent estimate of this.

 

[stopcrazypp]

Yes, I did, you just didn't understand it. The problem is temperature differential per volume per second. The radiator has an infinite supply of air at ambient temperature. The proposal is for an extremely limited amount of air being recirculated. You can always increase the flow over a radiator, if it starts to overheat, cooling is less per cubic foot of air, but more cubic feet overcomes this. With a fixed amount of air this doesn't work, since if you don't cool it down, it comes back hotter 4 seconds later.
So you'd believe me if I told I did the math (on the heat exchanger, I already showed you more math on the air flow temperature differential)? Didn't think so. Did he show you his math for the heat exchanger? No of course not. If he is capable of designing high performance heat exchanger why is he working as a financial journalist? He didn't even do the math I SHOWED you.

Fair point on the radiator overheating part, but the way you word your posts indicates you have not done the math, but rather just look at it qualitatively and say it is impossible. Note that the author is an engineer, not a financial journalist. Seekingalpha allows anyone to publish (you don't have to be a journalist).

We have done math on the CFM required to move the heat (which is not at impossible levels, so that supports the author's claims), but have not done the math on heat transfer radiator surface area. Again, with a large enough radiator that is sized enough to not overheat at the given heat load, then the author's idea would work. That is quite far from impossible.

Basically it would work the opposite (instead of removing heat from the coolant, it is tasked with transferring heat to the coolant, which removes it to the radiator outside). If those coolant pipes were swapped for refrigerant, then it works essentially the same as a refrigerator.

 

[Electroman]

ompare the number of molecules in a volume of gas vs a volume of liquid...
Also know that heat moves better via conduction through a solid or liquid faster than through a gas.

Take the other can and place it in a container full of water that's already been sitting in the fridge so it's the same temp as the air.

Thanks for mentioning these analogies - a soda-can immersed in cold water cools faster than a soda-can that only has cool air around. No disputing that.

But here is the issue. The battery cell (aka soda can) is NOT immersed in the coolant liquid (aka cold water). It only makes contact partially with a pipe that carries the coolant liquid. It is the lack of adequate thermal contact between those pipes and the cells makes a difference. If the pipes snake around every sq. inch of the cells then, ya sure. But it doesn't. Does that make enough difference to make air flow a better heat exchanger?. May be.

 

[lklundin]

A variation of the same problem exists in ICE cooling and as we all know...

I actually don't know much about ICE cooling. But I have reason to believe that it has less in common with the cooling of batteries than does the cooling of electronics: The batteries to be cooled are electrically conducting, unlike ICE components they cannot be subjected to very much pressure and their ideal working temperature is also a good deal lower. (Once warm, my ICE cooling water is steady at 90C, while batteries and electronics like it better around 30C - 40C. In addition to the water temperature my previous car (Audi A6 TDI) had also an engine oil thermometer, which during a good high-way drive could be steady at 170C, with the water at 90C).

So I think cooling ideas from IT may be more relevant here.

PS. Let me be the first here to state that I had to look up the Churchill–Bernstein equation.

 

[stopcrazypp]

Thanks for mentioning these analogies - a soda-can immersed in cold water cools faster than a soda-can that only has cool air around. No disputing that.

But here is the issue. The battery cell (aka soda can) is NOT immersed in the coolant liquid (aka cold water). It only makes contact partially with a pipe that carries the coolant liquid. It is the lack of adequate thermal contact between those pipes and the cells makes a difference. If the pipes snake around every sq. inch of the cells then, ya sure. But it doesn't. Does that make enough difference to make air flow a better heat exchanger?. May be.

From Tesla's patent filings and other pictures taken by those who opened the pack, the snaking metal pipe is in contact with roughly 1/4 to 1/6 of the circumference of the cell, is not the full height of the cylinder, and also has a ribbed thermal interface material around it that is non-conductive (to eliminate risk of shorting the cells).
Tesla or GM: Who has the best Battery Thermal Management?
Patent US20110212356 - Extruded and Ribbed Thermal Interface for use with a Battery Cooling System
Air cooled battery modules

 

[Swampgator]

Well, a typical CPU can generate 95 watts of heat output, concentrated in a 1.5 inch square. We use thermal compound, radiator fins, and ambient air to cool those typically.
What we need for each battery in a module is more like 2 watts/cell of heat removal, and Mr Carlson's proposal uses cold air (0C) so with enough airflow I fail to see the problem. He estimates the air will exit the module at 35-40C.

Here is another quote from his comments:

If cell internal resistance is as low as I believe is reasonable if Tesla supports Ludicrous mode, the power dissipation while SuperCharging will be 400W/module and the max cell can temperature will be 27 - 28C.

 

[Electroman]

From Tesla's patent filings and other pictures taken by those who opened the pack, the snaking metal pipe is in contact with roughly 1/4 of the circumference of the cell, is not the full height of the cylinder, and also has a ribbed thermal interface material around it that is non-conductive (to eliminate risk of shorting the cells).

Thank you.

In essence the arguments folks have been making that, 'liquids are denser and therefore has the capacity for better heat exchange', - all that is less relevant here. In the end a good thermal contact is essential, to exchange heat effectively.

You can snake a coolant pipe around a hot engine, covering 90% of its radiator fins. You can immerse uranium rods in heavy water. You can do none of those with a power cell.

 

[Swampgator]

Another possibility for Model 3 is that Tesla has copied BMW I3 cooling system, which using the A/C refrigerant under the 8 modules.
This may also be able to explain the larger pack size?

 

[stopcrazypp]

Another possibility for Model 3 is that Tesla has copied BMW I3 cooling system, which using the A/C refrigerant under the 8 modules.
This may also be able to explain the larger pack size?

I'm still skeptical of his claim of larger pack size. And the pack size can be explained by Tesla preparing for a larger capacity pack (akin to S/X capacities).

 

[J1mbo]

From Tesla's patent filings and other pictures taken by those who opened the pack, the snaking metal pipe is in contact with roughly 1/4 to 1/6 of the circumference of the cell, is not the full height of the cylinder, and also has a ribbed thermal interface material around it that is non-conductive (to eliminate risk of shorting the cells).
Tesla or GM: Who has the best Battery Thermal Management?
Patent US20110212356 - Extruded and Ribbed Thermal Interface for use with a Battery Cooling System
Air cooled battery modules

I think that patent was for the Roadster?

Tesla filed Patent 8968949 in 2015, which is (for me) a very strong indication that this is what they will use in M3, if not sooner.

A method for withdrawing heat from a battery pack is provided, wherein the heat is transferred from at least one electrode of each cell comprising the battery pack, via an electrically and thermally conductive tab, through a current collector plate and through a thermal interface layer to a temperature control panel that is coupled to an external temperature control system.

My understanding of this patent is that each electrode of the cell is connected to something like a bus bar, and a liquid-cooled, electrically isolated heat pipe sits on top of the bus bar.

 

[lklundin]

Well, a typical CPU can generate 95 watts of heat output, concentrated in a 1.5 inch square. We use thermal compound, radiator fins, and ambient air to cool those typically.
What we need for each battery in a module is more like 2 watts/cell of heat removal, and Mr Carlson's proposal uses cold air (0C) so with enough airflow I fail to see the problem. He estimates the air will exit the module at 35-40C.

When I compare the battery pack cooling to the cooling in IT, I would more think of an air cooled rack using maybe less powerful CPUs (many data centers use low-power CPUs like Intel Atom or also ARM now). But more importantly, the density of CPUs in the rack is much lower than the cells in a battery pack, so the air moving over the radiator fins you mention would be cooling a power per area comparable to the power per area that you mention for the battery cells. More or less. And the desired temperature would be pretty much the same.

So just from a cooling point of view, I also think an air cooled battery pack could work.

 

[diamond.g]

As a driver of a turbo charged car, I am seeing this as an air to air inter cooler vs an air to water inter cooler conversation. Air to water is better. We all know it. It also happens to be more expensive.

Now the OP is talking about refrigerating the air that is circulating around the batteries which is better than ambient and could even be seen as a reverse water (refrigerant) to air inter cooler.

 

[Yggdrasill]

I understand all that. But that has nothing to do with my point. My point is concerned about the amount of heat being able to removed from the inside of the pack using air.
Here are the interfaces:
1) Heat transfer from cell to air
2) Heat transfer from air to radiator inside pack
3) Heat transfer from radiator to coolant inside pack
4) Heat transfer from coolant to radiator in front of car (plus perhaps an AC/heating in between)
5) Heat transfer from radiator in front to air

#3-5 there is no dispute is possible, since that pretty much exactly like the S/X pack today. #2 I don't see how it is impossible, since it is basically the analog of #5.

#2 is the only one that is potentially problematic. And no, it isn't entirely analogous to #5. As you mention, this could be after an A/C unit, which means you may only need to reduce the temperature of the coolant from say 90C to 80C, with an unlimited supply of air at 20-40C. I don't think #2 is impossible, but my experience with heat exchangers tells me it's unlikely that a sufficiently capable heat exchanger will be sufficiently compact to be practical.

 

[lklundin]

Well, a typical CPU can generate 95 watts of heat output, concentrated in a 1.5 inch square.

As for air-cooling in IT, why are we even talking CPUs?

There are air-cooled, rack-mountable, off-the-shelf lithium-ion UPS-units with e.g. 12kW capacity (in 1U form factor), such as HP's R12000 DirectFlow 1U Rackmount UPS - three of those can be combined for

So there is no doubt that air cooling of high capacity lithium-ion battery packs is practical (apart from the obvious example, the PowerWall).

 

[diamond.g]

As for air-cooling in IT, why are we even talking CPUs?

There are air-cooled, rack-mountable, off-the-shelf lithium-ion UPS-units with e.g. 12kW capacity (in 1U form factor), such as HP's R12000 DirectFlow 1U Rackmount UPS - three of those can be combined for

So there is no doubt that air cooling of high capacity lithium-ion battery packs is practical (apart from the obvious example, the PowerWall).

Yeah in a chilled server room they work great.

Is the PowerWall not air cooled?

 

[lklundin]

Is the PowerWall not air cooled?

It is. Instead of 'Apart from' I should have written 'In addition to'. (I am aware that the PowerWall has design criteria somewhat different from a UPS (and a EV battery), e.g. a daily quite deep charge/discharge cycle. Still, its cooling has more in common with a module of an M3 battery pack than an ICE).