Science based discussion on charging LiFePO4 below 32°F

[eemarty]

For the typical thick electrode prismatic 'blue' cell used by DIY'er, should not be charging above 0.5 C(A) at any temperature.

Very interesting. Many manufacturers support 1.0C "maximum" charge currents. Several of these have lower "recommended" charge currents of 20A, 50A, or 63A. The Jakiper is an outlier here and even supports 1.0C at -5°C though it's possible that their BMS is doing some smart things and reducing charge current at lower temperatures.

It wouldn't be hard for an engineer designing a BMS to scale charge current with temperature and state of charge. Seems like it could be done in firmware. For example: 10A@-10°C, 15A@-5°C, 20A@0°C, 30A@5°C, 40A@10°C, 100A@20°C. This would allow them to better protect the battery from damage at cold temperatures. But I'm not sure if any of the manufacturers have figured that out yet or implemented it.

EG4 Battery: CAPACITY: 100AH, MAXIMUM CHARGE CURRENT: 100A CONSTANT, CHARGE RANGE: 32 F - 113 F
Jakiper: Nominal Capacity: 100Ah, Maximum Charge Current: 100 A, Charge Temperature: 23 ~ 131 °F ( – 5 ~ 55 °C )
SOK: Nominal Capacity: 100Ah, Maximum Charge Current (Continuous): 100A, Charge Temperature: 32°F – 113°F
Trophy: Rated Capacity (AH): 100 Amp-Hours, Continuous Charging Current (A) 100 Amps, Cell Charging Operating Temperature Range: -0～45 ℃

 

[MisterSandals]

it's possible that their BMS is doing some smart things and reducing charge current at lower temperatures

I am not sure if any BMS has more than 2 states, on and off.
Besides…

It wouldn't be hard for an engineer designing a BMS to scale charge current with temperature and state of charge.

Isn’t this what charge controllers are designed to do (scale current):

I am not sure I would want a more complex BMS as it would likely reduce reliability. And for the last line of defence, BMSs need more reliability if anything.

 

[eemarty]

I am not sure if any BMS has more than 2 states, on and off.

Yeah I guess the charge controller would be more suited to reduce the current under low temp conditions. But if that's the case, why do we expect BMS's to have low temp protection. One issue is that the temperature should be monitored close to the cells themselves. The cells may be at different temperature than ambient due to self heating so communication is needed between the BMS and the charge controller. Another issue is that state of charge should also be monitored by the charge controller. It seems like recent server rack batteries have this communication.

 

[time2roll]

Isn’t this what charge controllers are designed to do (scale current):

I am not sure I would want a more complex BMS as it would likely reduce reliability. And for the last line of defence, BMSs need more reliability if anything.

Certainly in an electric vehicle the charging rate is actively adjusted based on battery temperature.
Not aware of anything off the shelf that does this.

 

[MisterSandals]

Certainly in an electric vehicle the charging rate is actively adjusted based on battery temperature.

Agreed but my point is that its almost certainly the charge controller doing this, not the BMS.

Yeah I guess the charge controller would be more suited to reduce the current under low temp conditions. But if that's the case, why do we expect BMS's to have low temp protection.

BMSs provide several different kinds of protection. Why would we NOT expect our last line of defense for our battery to protect it from low temp charging?

 

[curiouscarbon]

I am not sure if any BMS has more than 2 states, on and off.
Besides…

Isn’t this what charge controllers are designed to do (scale current):

I am not sure I would want a more complex BMS as it would likely reduce reliability. And for the last line of defence, BMSs need more reliability if anything.

the basic feature of safety disconnect is just the beginning

most cheap BMS do not share data, much less provide instructions for other devices.

in the electric vehicle i drive, the BMS handles safety disconnect as well as instructing charge controller and battery heater/cooler.

 

[eemarty]

Why would we NOT expect our last line of defense for our battery to protect it from low temp charging?

Well I can think of 2 reasons:

1) There are many people who would like the ability to charge at low charge rates at slightly below freezing temperatures. A hard cutoff at 32°F in the BMS prevents this from being possible.
2) Giving the user a hard cutoff at 32°F gives a false sense of security that high charge rates at 33F are safe. As far as I can tell they probably are not.

I'm not saying low temp protection in a BMS is a bad thing, but there are tradeoffs involved. It probably prevents lots of people from damaging their batteries, but it also has the drawbacks mentioned above. A smarter system would have the BMS tell the charge controller what the maximum charge rate is based on the temperature of the cells and the current state of charge.

Actually, thinking about it again, the BMS could allow different charge rates at different temperatures even if it didn't communicate with the Charge Controller. It could allow 0.1C@-10°C for example. If it sensed that the allowed charge rate was exceeded, is could just shut off current and go into protect and then you'd be in the same boat as if you had a hard cutoff at 32°F. But if you used a low enough charge rate to be safe, the current would be allowed to flow.

 

[time2roll]

Temperature compensation needs a new function.

 

[sunshine_eggo]

FWIW, my Batrium/Victron combo allows a couple different options.

1. Completely turn off the MPPT based on temp (currently set to below 3°C).
2. Disallow charging but allow the PV to continue to power loads (Between 3-8°C).

I chose temperature values based on a perceived measurement with the batrium sensors. I have a total of 3 Batrium sensors and 4 Victron sensors (SBS, BMV and 2X inverter probres) as well as an IR gun I use to spot check. My best guesstimate is that the Batrium sensors read 3°C low.

Initially, option 2 was allowing some bleed through, but I think this was more of a GX issue. It would allow anywhere from 0.2-0.7A of charging even when charging was set to 0A either by the BMS or by the GX. After a few days, the behavior has corrected itself allowing a typical -0.3A discharge, which is consistent with prior charge regulation on my old FLA (30A limit saw a typical 29.7A input).

Rapid load changes do cause fraction-of-a-second transients that do send some current to the batteries.

If the GX continues to display this consistently, I will likely lower that threshold to permit PV to power loads down to -20°C, which is colder than it's ever been in the area.

The above demonstrates the value of having BMS communication with the charging hardware.

 

[eemarty]

I found another paper which tests various charge rates at -10°C and specifically uses LFP batteries. Again the full text isn't public, but I have the PDF so message me if you would like a copy.

I'll quote from the paper:

Based on the test results, when charging a LiFePO4 battery in a low temperature environment, here -10°C the charge current rate should be restricted to less than 0.25 C and the cut-off voltage to less than 3.55 V.

Low temperature cycle life experiments were performed at -10°C, and quantitative methods were used to identify the LFP battery aging mechanism. Capacity fade was more severe with a higher charging rate and higher cut-off voltage, and the turning points for the current rate and cut-off voltage after which degradation accelerated were determined to be approximately 0.25 C and 3.55 V.

These are the test variables:


All these results are after 40 cycles at -10°C. The key chart is here:


One thing that's a little different with this paper is that they test various cut off voltages for the final constant voltage (CV) portion of the charge and find that lower voltages reduce the capacity degradation. The CV is usually user configurable on charge controllers and is often set to 3.65V per cell (14.6V for a 4s, 12.8V system). This paper found that keeping the cutoff voltage below 3.58V drastically reduced capacity loss.


And also they tested several cells at 3.65V in the CV part of the CC-CV charge but they varied how long the CV portion lasted. The charge ended when the current dropped below 0.5C, 0.1C or 0.05C. They found ending the charge sooner reduced the degradation even at a 0.5C charge rate at -10°C!


Both of these results suggest that trying to get that last 1% capacity into the battery at very cold temperatures can cause damage. These are settings that we can usually configure on our charge controllers.

 

[sunshine_eggo]

How did cell 8 fare?

If it hasn't been mentioned already, all of these tests seem to be a relatively few number of cycles relative to the quoted life of LFP. Using a very conservative 2000 cycle life to 80% SoH, that's 0.0005% average degradation per cycle or 0.025% per 50 cycles. In other words, cell #4 is deteriorating 163X faster than the quoted cycle life.

IMHO, any test results should be compared to the .0005%/cycle degradation value.

I'd really like to see 0.1C to 3.45V with 0.1C cut off.

 

[eemarty]

How did cell 8 fare?

Strangely they didn't list the results of each cell in a table format, but you can see that cell 8 was the only one that used 0.1C charge rate so it must be the leftmost square on the charge rate curve. My eyes see that as about 2% capacity loss.

If it hasn't been mentioned already, all of these tests seem to be a relatively few number of cycles relative to the quoted life of LFP. Using a very conservative 2000 cycle life to 80% SoH, that's 0.0005% average degradation per cycle or 0.025% per 50 cycles. In other words, cell #4 is deteriorating 163X faster than the quoted cycle life.

IMHO, any test results should be compared to the .0005%/cycle degradation value.

Agreed, it would have been nice if they had at least included some less harsh conditions in their test program like a few data points at 0°C and 10°C. These tests must take weeks to run and require expensive equipment. It makes you wonder how much degradation testing the manufacturers do before they come up with their charging guidelines or if they simply rely on this type of academic paper too. If they are doing their own degradation tests at harsh conditions, I wish we could see that data, but it's probably kept secret for competitive reasons.

I'd really like to see 0.1C to 3.45V with 0.1C cut off.

That would be similar to Cell 8, but with lower CV voltage. Cell 8 had about 2% capacity loss.
It would also be similar to Cell 11, but with lower CV voltage and much lower charge rate. Cell 11 was the only cell with 3.5V cutoff voltage so you can find the point above 3.5V on the cutoff voltage chart. It looks about 1% to me.
I would expect degradation at the conditions you suggest above to be less than either of these.

 

[Old_Skewler]

This is the key chart for our purposes:

View attachment 120815

Very informative discussion.

This is my very first winter off-grid running on LFP batteries and I had to adjust behavior to accommodate battery limitation in cold temperatures.

 

[fastline]

I am curious of you guy's setup that is allowing batteries to dip to these temps? Are these mobile RV applications? At least around here, I almost always recommend an underground vault. If not that, a super insulated cell around the batteries with a small resistive heater.

In anycase, as luck would have it, I actually have to push to get some LFP packed slightly charged tomorrow. They are stuck in a semi trailer, no heat, and on a pallet. I can't get them to a heat source. I was going to charge the other day but realized with my FLIR camera they were about 15F. I checked today and they were right at 32F. Ambient temp will be way over that but I have no way to let the sun heat them.

I will only be pushing about 400W to these 5kwh packs. Like a trickle! Curious you guy's thoughts on this? On the off chance internal temps are still a touch low? This is a one time thing just to get some to a safer charge level before the real cold sets in. They will get installed in the next few mo anyway. I just can't stand to have voltages down in the 2.7V range. I want to see 3.2V on them is all.

 

[sunshine_eggo]

40' uninsulated shipping container

I will be building an insulated battery box.

I also have a hairbrained scheme to provide limited electrical heating to a portion of the shipping container...

Personally, I wouldn't try to charge them @ 15°F at any level. I would consider breaking down the pallet and getting them into the sun one at a time.

The bigger question is how did they get to 2.7V? Cells should almost never be at that voltage. If you received them like this, I would seriously reconsider deploying them without in-depth testing.

 

[MisterSandals]

I am curious of you guy's setup that is allowing batteries to dip to these temps?

My gut feel is that its mostly your typical cryogenic lab, garden shed or shipping container.

 

[fastline]

40' uninsulated shipping container

I will be building an insulated battery box.

I also have a hairbrained scheme to provide limited electrical heating to a portion of the shipping container...

Personally, I wouldn't try to charge them @ 15°F at any level. I would consider breaking down the pallet and getting them into the sun one at a time.

The bigger question is how did they get to 2.7V? Cells should almost never be at that voltage. If you received them like this, I would seriously reconsider deploying them without in-depth testing.

I didn't attempt any charging yet. Most batteries were reading 32-38F today, and tomorrow will be even warmer with no freeze temps for the next few days. So I am pretty curious how "set in stone" this 32F rule is? Obviously I don't want to inflict damage trying to help them.

 

[sunshine_eggo]

I didn't attempt any charging yet. Most batteries were reading 32-38F today, and tomorrow will be even warmer with no freeze temps for the next few days. So I am pretty curious how "set in stone" this 32F rule is? Obviously I don't want to inflict damage trying to help them.

This thread has presented pretty compelling evidence that it's a real thing. It's not binary, but there's a range. No data has been presented that you could point to that would allow you to say:

"I can safely charge my X Ah bank at 0.YC Amps below freezing and everything will be fine."

In reality, there is some evidence that charging around freezing can also be detrimental. The issue will also vary from manufacturer to manufacturer, and no data has been traceable to commonly used cells on this forum.

I'm personally waiting until my cells get to 8°C (46.4°F) before I hit them with 0.1C. Given potential measurement error, that might be as low as 41°F.

My concerns about this issue arose from my hardware allowing 0.2-0.7A (0.0016C) of charge to my 450Ah battery bank, and I'm still not confident that's completely safe. Sure, a few times, I'm probably fine, but every morning for the next 3-4 months? Nope.

Again, the bigger question is how the hell did these cells get to 2.7V?

 

[eemarty]

Everyone is free to read the studies in this thread and come to their own conclusions. I doubt we will ever get to a point where everyone agrees on what is "safe" and that's totally fine. It's also important to know that your cells are degrading all the time. There's no getting around it. We all get old and wear out, and so do our batteries. Some manufacturers may have warranties that you might take into consideration. Some manufacturers may have data about low temp charging available if you ask. Some manufacturers have guidelines that are much more aggressive than what I'm suggesting below. Jakiper allows a 1.0C charge at -5°C.

My personal opinion, based on what I've read, is this:

Ideally, I would keep my cells at 10°C to 20°C (50°F-68°F) all the time. This seems to be the sweet spot between low temperature degradation and high temperature degradation. If that wasn't practical, I would limit charging like this:

Below this temperature:	Limit charge current to:	Stop charging when cell voltage reaches:
10°C (50°F)	0.4C	3.60V
5°C (41°F)	0.3C	3.55V
0°C (32°F)	0.2C	3.50V
-5°C (23°F)	0.1C	3.50V
-10°C (14°F)	0.05C	3.50V
-15°C (5°F)	don't charge	don't charge

 

[eemarty]

Well I just came across this manufacturer that has a table of allowed charge current vs temperature for a 50Ah LFP cell. They have several cells in their range that are rated to charge down to -10°C. Then there are several cells which are only rated to charge down to 0°C. So I'm not sure if that difference is due to an inherent difference in the cell itself or just that the different products are tested and qualified to different conditions. Either way, I've never seen this much detail in a manufacturers charging spec.



They also give some interesting details on state of charge vs temperature. This battery seems geared for the automotive market where regen braking provides very high energy for a short time.