Living in the Free State (RSA), we regularly experience sub-zero temperatures over night during the winter. It is known that lithium batteries do not perform well when cold, and can be damaged if charged when below freezing temperature. My inverter and 2 x 10kWh LiFePO4 batteries are housed in my unheated garage, and I've been quite concerned that the batteries could become too cold to operate optimally. After some trawling of the internet, I concluded that it would be best to isolate them in a "box" of polystyrene sheeting, and to add two small (7W each) heating mats below the batteries to add a small amount of heat to this "box".

I initially used some duct tape to just stick the panels together as a "proof of concept" pilot project, and then added the heating pads, purchased from a local pet shop (sold as reptile heating pads...) after the first week of operation. After a second week of careful monitoring, being concerned now to not over-heat the batteries, I proceeded to properly build a box using 2x25mm thick polystyrene sheets as walls, and an insulating "floor" made from fibre-cement ceiling board mounted on a 6mm plywood sheet, and raised off the concrete floor with some 38mm x 38mm pine brandering for the two heating pads (placed in shallow wooden boxes and covered with a thick layer of kitty litter to act as a thermal mass below the batteries).

Here are some photos.

The un-insulated batteries, showing the insulating "floor" and heating pad boxes with kitty litter...

The duct taped temporary box...

...And the final enclosure before final fixing to the wall

So far, temperature variation during the last week ranged from about 14 degrees C minimum to 17 degrees C maximum, even though we experienced a few nights of zero degree C minimum air temperatures in the early mornings. We expect some typical -4 to -6 C lows in the next week or two, and I will report back on performance after the next cold fronts.

I am quite pleased with the outcome of this little project.

Your feedback, comments, and constructive criticism will be appreciated.

Edited September 9, 2025Sep 9 by HennieL

32 minutes ago, HennieL said:

-4 to -6 C

Shure, the outside temperature goes that low just before sunrise. After multiple years in die Free States and those lows, I haven't seen freezing temperatures inside a garage, nor did my BSL have issues.

Discharging is fine down to -20 degrees C... Why not just cycle the battery to heat the cells?

42 minutes ago, frivan said:

Discharging is fine down to -20 degrees C... Why not just cycle the battery to heat the cells?

Well, for one thing, I already extract about 2kWh from the batteries every morning before sunrise to heat the geyser. Discharging more than this some 2-3 hours before charging can begin via PV panels would drain the batteries further than I would like, so that's not an option.

Even though discharging may be possible down to -20 degrees C, it does not mean that it would be good for the batteries - after all, power is supplied as a result of a chemical reaction, and as most people would know, chemical reactions happen better/easier/quicker at optimum temperatures - which for LiFePO4 happen to be quite a bit higher than your quoted -20 degrees C... so discharging might be "possible", but arguably not "fine".

Anyway, I prefer to do things as best I can, and that's my choice - further CONSTRUCTIVE feedback is still welcomed...

Just to elaborate further regarding the "fine" temperature ranges for LiFePO4 batteries - Summarized by my good friend ChatGPT:

True Electrochemical Optimum for LiFePO₄

Based on lab-tested data and peer-reviewed sources:

Parameter	Optimal Range	Notes
Charge Efficiency	15°C – 25°C	Peak coulombic and energy efficiency
Discharge Efficiency	10°C – 30°C	Capacity is near-maximal with low resistance
Degradation Minimization	15°C – 25°C	Lower electrolyte oxidation, minimal lithium plating
Internal Resistance (Low)	20°C – 25°C	IR is lowest in this band

Supporting Technical Sources

1. LiFePO₄ Cell Behavior at Different Temperatures – Detchko Pavlov et al.

Shows optimal ionic conductivity and lowest cell resistance between 20°C and 25°C. Performance degradation and voltage instability observed below 10°C and above 35°C.

2. “Degradation Mechanisms in LFP Batteries” – Journal of Power Sources (Elsevier, 2017)

Identifies thermal sweet spot for LiFePO₄ between 15–25°C for long-term stability and maximum energy throughput.

3. Battery University & Victron Energy (practical integration guides)

Both recommend targeting 20–25°C for best compromise between efficiency and lifespan — especially for systems cycled daily (like solar).

4. EVE 280Ah LFP Cell Datasheet (widely used in Sunsynk and similar packs)

Peak test cell performance (capacity retention over 6000+ cycles) occurs at 25°C constant ambient.

Have you tested these 7w pads make? I have a feeling you will have to go a bit larger to actually make a difference in the space.

In our turtle enclosure, we use a 7w pad to heat the hideaway, but in the morning it struggles to raise the temps, this is only about 4cm away from the temp probe, which is directly above a plastic enclosure.

PS: In battery packs that have heating, the pads are directly under the prismatic cells.

For further control, connect the pad to a sonoff switch, which can be turned on_off from something like homeassistant or nodered.

Edited June 19, 20251 yr by Pho3niX90

12 hours ago, HennieL said:

Just to elaborate further regarding the "fine" temperature ranges for LiFePO4 batteries - Summarized by my good friend ChatGPT:

He ain't your friend... btw, but really what was the cell temperatures before you went and did the acrobatics with boxing in the batteries and adding the heating pads and how do the cell temperatures compare now... the temperatures outside the enclosure do not mean much, I'd imagine...

@HennieL

Nice project. Just for curiosity sake, how does the temps differ from before to after?

Here's how I measure:

Yeah, a bit over the top, but rather safe than sorry. I'm watching the cell delta voltages like a hawk, and get WA notifications when Temps exceed 30 Deg or Delta Voltage >150mW

9 hours ago, Pho3niX90 said:

Have you tested these 7w pads make? I have a feeling you will have to go a bit larger to actually make a difference in the space.

In our turtle enclosure, we use a 7w pad to heat the hideaway, but in the morning it struggles to raise the temps, this is only about 4cm away from the temp probe, which is directly above a plastic enclosure.

I checked them in the open, and inside my study, and each unit reached a maximum temperature of 34`C - 35`C after about 30 minutes, at an ambient temperature of 18`C, and their temperature controllers set to about 80% of maximum. After installing them below the batteries I slowly ramped up the temperature during the first week to about 50% of maximum, being concerned to not over heat my batteries. After completion of construction I further increased the settings to about 75% of maximum, whilst the overnight minimum temp dropped to around zero, and the battery temperatures still increased slightly, currently giving me a BMS temperature of minimum = 15.9`C and maximum = 17.0`C for the last 24 hour period, with a minimum overnight temperature of -2`C reported by my Azure controller.

7 hours ago, Sidewinder said:

@HennieL

Nice project. Just for curiosity sake, how does the temps differ from before to after?

Thanks Sidewinder. I cannot give you a meaningful figure, as ambient temperature was quite high when I started, and dropped quite substantially during the past two weeks - the "before starting" BMS temperatures were definitely higher before starting with this project than they are now, but I believe the temperatures would have been much lower now if I had not completed this project just in time for serious winter 😁

I suppose I could switch off the pads for one night to measure their actual effect, but the insulated box would still skew the readings, unless I also break down the box again - and that's not happening until mid spring...

By the way, I don't think you're over the top to be concerned about your battery temperatures - or we both are, but then that's also OK. Given the cost of batteries, and the fact that both charging and discharging then outside of the optimum temperature range will shorten their operational life span, I would rather spend a few Rand and a weekend of my time to ensure that they ARE in fact operating within the Goldilocks temp range.

8 hours ago, Kalahari Meerkat said:

He ain't your friend... btw, but really what was the cell temperatures before you went and did the acrobatics with boxing in the batteries and adding the heating pads and how do the cell temperatures compare now... the temperatures outside the enclosure do not mean much, I'd imagine...

Well, "he" is not really my friend - that was tongue in the cheek - but I do use AI regularly in my work to crunch large data sets, and compile statistical analyses of this data... save me from having to do so myself. I do, however, do my own checks, and have learned not to trust everything presented - in fact, I probably have to correct it 2-3 times per session, but that's what I'm paid for. Actually scary what AI can do, and it's still in it's infancy 🫠 Anyway, let's not dwell there.

As stated in my post above, I make use of the internal temperature reported by the BMS to the inverter every few minutes - can't tell you if it's an average of individual cell temps, though...

23 hours ago, HennieL said:

Well, for one thing, I already extract about 2kWh from the batteries every morning before sunrise to heat the geyser.

I would guess that 1% losses is possible during discharge. For a 3kW geyser that would mean 30W inside the batteries, with better positional efficiency than a heating pad on the steel box. Does someone have information on cell temperature rise during discharge?

I am sure there are lots of losses in the inverter, why not put the batteries and inverter in a box?

16 minutes ago, frivan said:

I would guess that 1% losses is possible during discharge. For a 3kW geyser that would mean 30W inside the batteries, with better positional efficiency than a heating pad on the steel box. Does someone have information on cell temperature rise during discharge?

I am sure there are lots of losses in the inverter, why not put the batteries and inverter in a box?

Inverters are already running high in summer so all instructions for positioning indicates some free air around all sides.

What is good in winter is bad in summer.

2 minutes ago, Scorp007 said:

all instructions for positioning indicates some free air around all sides.

I agree that cooling in summer is required: I believe in a temperature controlled extractor fan.

47 minutes ago, frivan said:

I would guess that 1% losses is possible during discharge. For a 3kW geyser that would mean 30W inside the batteries, with better positional efficiency than a heating pad on the steel box. Does someone have information on cell temperature rise during discharge?

I am sure there are lots of losses in the inverter, why not put the batteries and inverter in a box?

Well, I can confirm that my 2kWh early morning extraction from the batteries increase the BMS temperature (inside the slightly heated and insulated box...) by between 0.2 and 0.3 degrees C

The batteries heat up by about 1.1 - 1.3 degrees C when charging roughly 6kWh from about 07:30 to about 11:00 in the mornings, charging at roughly 50A - thus roughly double as effective in heating the batteries than discharging them.

I thought that I should share some stats, for those who are interested. Below are the average minimum and maximum temperatures and standard deviations (SD's) of the batteries, based on temps reported by the BMS, for the various periods while this box was in it's various stages of "development". As can be seen, the SD's (i.e. variability of temperature...) decreased by more than 3 fold since starting with this project. For the last data set, the "Complete mounted" means that the "box" walls are now tied to the garage wall with brackets, closing up some rather large gaps where the panels previously pulled away from the wall.

Edited June 21, 20251 yr by HennieL

Correct me if I am wrong: battery temperature was never below 10 degrees Celsius.

Correct - because initial readings were in a VERY HOT early Autumn, and the insulating box and heaters were operational when real winter finally hit.

Your point is ???

52 minutes ago, HennieL said:

Correct - because initial readings were in a VERY HOT early Autumn, and the insulating box and heaters were operational when real winter finally hit.

Your point is ???

I haven't seen the batteries going out of your chat GPT optimal discharge temperature range. Your method lacks science: at what temperatures do you need how much heating.

5 minutes ago, frivan said:

I haven't seen the batteries going out of your chat GPT optimal discharge temperature range.

May I say "I rest my case". It's quite obvious that the insulated box idea that you appear to dislike is in fact working very well, with diminishing standard deviations giving an ample statistical confirmation of this.

31 minutes ago, frivan said:

Your method lacks science:

And you sir, lack manners.

I'm sharing rather detailed, and well presented empirical data, with the hope that it might be of interest to fellow Forum members. I am NOT presenting a dissertation for a PHD - if I did, you can be assured that it would have contained more than enough science. If you really want more "science" I will be willing to do a really detailed study for you over the next year (at least, to cover variables such as changing seasons), but then you will have to pay me my professional fee - and I'm not cheap 🤑

So, let's rather not quibble about this any further, and agree to disagree on everything...

I do exactly the same as @HennieL even though I live in a more temperate climate (Stellenbosch). I use a box insulated with 50mm Isoboard and heated with 2 28W heating pads (Adjustable Heating Pad With Plug 28W | Shop Today. Get it Tomorrow! | takealot.com). In summer the box gets cooled with a small air conditioner when the batteries approach 30C. I always keep the cell temperatures between 20 and 30C.

The motivation is simple: Pylontech's own research shows the huge impact of temperature on cycle life.

12 hours ago, HennieL said:

pay me my professional fee - and I'm not cheap 🤑

So, let's rather not quibble about this any further, and agree to disagree on everything...

IDOEW has been published by ECSA, I will subtract my fee.

Just add the ambient temperature and do a few tests without the pads switched on. That would pin down some significant variables.

SD discussions sounds like people reloading for 6.5 Creedmore.

7 hours ago, frivan said:

SD discussions sounds like people reloading for 6.5 Creedmore.

I hope you're not one of them... Slow and heavy don't need SD to be able to bring the meat home 😁😁

7 hours ago, frivan said:

... do a few tests without the pads switched on. That would pin down some significant variables.

Not going to happen while it's so cold... let's talk again in a few months.

On 2025/06/21 at 4:45 PM, HennieL said:

I thought that I should share some stats, for those who are interested. Below are the average minimum and maximum temperatures and standard deviations (SD's) of the batteries, based on temps reported by the BMS, for the various periods while this box was in it's various stages of "development". As can be seen, the SD's (i.e. variability of temperature...) decreased by more than 3 fold since starting with this project. For the last data set, the "Complete mounted" means that the "box" walls are now tied to the garage wall with brackets, closing up some rather large gaps where the panels previously pulled away from the wall.

@HennieL I would like to thank you for valuable information shared with this forum on your thoughts mitigating charge/discharge under sub zero tempratures.

10 hours ago, frivan said:

Just add the ambient temperature and do a few tests

I do not believe that outside ambient has much to do in the data supplied in the table. The important factor is keeping the battery heat constant inside between charge/ discharge cycles.( keeping cold out is not the same as keeping heat in) For that you need effective multi layer insulation with good R(resistance to heat flow)-value like extruded polystyrene and or external heating pads with temp control.

11 hours ago, frivan said:

SD discussions sounds like people reloading for 6.5 Creedmore.

This implies flat standard deviation within the data set that is compared to ammunition round with a flat trajectory. Let us respect in which the OP presented his findings on "keeping heat constant within a lifep04 battery during charge/discharge cycles in sub zero tempratures.

Thanks @TaliaB , appreciate your comment.

Here's one last graph to keep @frivan occupied... Although not 100% mathematically correct, the error bars (shaded areas) in the graph below gives a close approximation of the standard deviation for the various periods. I have also added the maximum and minimum air temperatures that you requested.