LiIon 18650 @ -29°C for 72h

[harerod]

I was asked this question yesterday: "How would a Panasonic NCR18650GA react to being stored once at -29°C for 72h?"
After a look at the datasheet my immediate formal answer was: "This is outside the spec. So either get clearance from the manufacturer, get another cell, or change the rules of the test."
 
After reading a bit further into the topic, I found that -20°C seems to be a typical limit for operation and storage of LiIon cells. Below that temperature the cell chemistry seems to change significantly. I also found LiIon cells that are specified for -30°C.
 
Out of curiosity and since I can't do -29°C in my lab: could anybody share actual experience with cold LiIon's, maybe even a NCR18650GA 3500mAh? How would that cell react to simply being "frozen" and "thawed"? How would that cell react to a light load (e.g. 10mA) in the "frozen" state?

[Siwastaja]

Absolutely fine. Max safe charging current goes to so close to zero that it's a very good idea not to try charging at all; also because manufacturers often specify max charge current as a simple step function forbidding it completely below 0degC or -5degC. (If you engage in a big project with the cell manufacturer with NDA and stuff, then they would come with actual curve sets on allowable charge currents, which would be non-zero even at very cold temperatures.)

Discharging is fine; DC ESR skyrockets, but that isn't a problem for a 10mA load. Larger loads cause more self-heating which in itself isn't dangerous but to be on the safe side you might want to avoid doing that; maybe too quick heating could cause mechanical issues within cells. Hotspots and uneven thermal expansion, the usual stuff. But 10mA definitely isn't a problem. If it were my project, I would consider limiting myself to 500mA or so, also for efficiency and ability to fully discharge; if I need more current than that, then parallel more cells. You see, in cold the terminal voltage sags more and last thing you want is having to stop because of reaching cut-off limit already at 50%SoC.

Same and similar cells are used below the -20degC datasheet ratings; but li-ion batteries being specialty components, each customer gets their own datasheet. It's only us small players who need to work with generic datasheets (usually with pretty limited information).

For an interesting reference, I drove my Nissan Leaf, which does not have battery heating*, which had sat outdoors at between -25..-30 for a few days; the BMS limited discharge power from the usual 160kW (2.6C for the 62kWh pack) to 70kW or something (1.1C). Take this with a grain of salt, but this gives you a rough idea how manufacturers who give long warranties on their batteries treat their cells in extreme conditions.

*) the smaller 40kWh version does have a small pack heater

[harerod]

Hi Siwastaja,
I actually planned to address that post "Att: Our resident Finns", but forgot. I was hoping that you could share some insight. Sadly, I lost my contacts to some nice rather nice Sami in Karasjok (Norway) - but still remember cool winter stories shared twenty years ago on a hot (+7°C) June day during barbecue.
 Coming back to the cells: Eventually my client will need well over 100k units, therefore I sent them to the manufacturer straight away. I just made sure that my electronic design would be safe, critical spot being some Panasonic elcos, rated -40°C.
Again - for a commercial medical device I would make sure to stay well within the datasheet specs. Thank you for sharing your experience.
 What prompted my question were hints in the abstract of some Chinese paper and  https://en.wikipedia.org/wiki/Lithium-ion_battery that could be interpreted that the cell chemistry changes and other chemical processes take place. That got me interested in asking for actual experience.
 A Leaf using 160kW? That's the Finnish version with 100kW for heating, I assume? My wife drives a Toyota Hybrid, which I am sure has been cold tested on some frozen lake north of the polar circle. Since I have seen -30°C rated cells, I wondered if those might use a different solvent. That tiny battery pack surely isn't heated.

[Siwastaja]

100k units is such large number that working with the manufacturer becomes possible. And of course, if it's a medical device then even more so. Product data sheets for stuff like li-ion cells are not set in stone, but rather the manufacturer might well qualify that very same part for use down to -40degC as long as they can micromanage the details how the cells are actually used.

Those "other chemical processes" would mostly mean metallic lithium plating at the negative (graphite) electrode, but this only happens in charging, which is why maximum allowable charge rates are so important and complex functions of temperature, state-of-charge and cell aging. For example, while said Nissan Leaf pack can still deliver said 70kW discharge at near -30degC, charge would be limited to something as ridiculously low as 5-10kW or so (I haven't tried quick charging in such conditions. I charge at 4.5kW at home and this is always below the BMS limit.)

But on discharge side, I don't think there are other chemical reactions. Extreme discharge rates would cause physical problems (quick thermal expansion, hotspotting). AFAIK the electrolytes (lithium salt in organic solvent) do not contain anything that suffers at -30degC or even at -40.

160kW which is 210hp or so is not that much for a car. Look at Tesla's performance models for even more power. Or many conventional combustion engine cars. Cabin heating at -30degC consumes 4kW or so, which is quite significant in range reduction, but battery efficiency degradation (aka the skyrocketing DC ESR I mentioned) also becomes a big loss element - I don't have a heated garage which would help tremendously. Combination of these effects halves the driving range compared to summer. More modern EVs that have well-designed battery heating and better thermal insulation on the pack score better and lose less range.

[harerod]

Thank you for following up on this. Especially for emphasizing the electrochemical issues during charging.
 
As so often with long running projects, you need the most support during development, while the large volume may be a couple of years or further down the road. From what I hear, getting actual support, instead of well worded nothings, seems to be a bit of an issue.
There are several intermediate solutions available, though. One involves defining a -20°C limit during transport and storage. The other would involve COTS cells that are rated to -30°C. During design I recommended using four single 18650 in THT mounted clips initially. Compared to a cell pack, we will have the flexibility of a broad COTS supply. The moment production finally ramps up, we can always switch to packs. The first version of that device used a TI BQ40 battery manager. That proved to be an issue regarding abysmal support by TI and massive sourcing issues during Chipageddon (€16 instead of €2). For the current revision I designed a cell manager from scratch, essentially using some control pins on an existing MCU and adding some discretes. We have the PCB space and this solution will actually be cheaper regarding components.
 
Regarding your car: I had no clue. When I read Leaf, I imagined somathing like my friend's Dacia Spring (33kW / 27.4kWh (I guess the .4 is significant)).

[Siwastaja]

I guess that a cell manufacturer (or anyone analyzing your design for safety) will have more to say about using 18650 holders, than using cells below -20degC. After all, these cells are meant to be spot welded into a pack, and not put in holder. Some holders I have seen impose a real risk of puncturing the thin plastic wrap around the cell, shorting the cell case (negative) into the positive tab during insertion. I have only seen these holders in amateur projects, and used them as such myself in hobby projects, but I would be at least very careful with them in an actual product.

But maybe it's acceptable for a smaller batch, like prototypes to selected customers, or if only assembled by sites you can control. An instruction leaflet along the lines of "first push the positive end into the holder, then push the negative end in place" could mitigate the puncture risk.

[harerod]

Yes, the holders require additional attention in cell selection and mechanical setup. We got our design certified. Even during normal operation, the holders also add more impedance than all the PCB traces and switches. This works for our specific charge/discharge requirements. The pack current goes through a 2.5A UL approved fuse, top operational current is well below that.
 
Edit: The only excuse for using the holders is the unloading of the procurement process for the final cells to a later point. When a pack gets installed eventually, it will require no electrical redesign. However, the clips will be replaced by whatever connector that pack manufacturer will propose. Or the pack's wires will be directly soldered in. Whatever works best at that time.