On Keeping LFP Warm

[fafrd]

Freudian slip? My daughter lives/works in Santa Rosa. I'm going to have to start telling her she lives in/near the "bat area".

Probably . They can be pretty ba

I think your heat loss calculation is off a bit. I'd need to double check this, but on first blush: You are using the American "R" value which doesn't have units that get you to Watt hours.

I got my definition:

Rival in (K⋅m2/W) is the R-value

from here: https://en.m.wikipedia.org/wiki/R-value_(insulation)


If that’s not the correct R Val definition to use with the foam insulation we purchase at Home Depot or for soft wood, I appreciate an education.

You want to use the metric "RSI" value, which is simply R / 5.678. So your RSI value would be 1.06/5.678 = 0.1866854 °C*m²/Watt. Dividing your 10°C differential by this RSI value gives you 10/0.1866854 = 53.566 W/m². Multiply that by your area of 0.76m², and you get 40.7W, or 977Wh/day. That seems high to me, so maybe my calculation is wrong too!


You are right that increasing the R value 6 by using foam will decrease the Wh by the ratio 1.06/6, so 40.7W would go down to 7.19W, or 172Wh/day.

You say "cellar temperature" is constant 65°F, but from all the rest I assume you are NOT putting your batteries below grade (in the cellar). If you are, you have no calculating to do!

Our cellars here in the Bat Area tend to be above grade. So the battery will be in an under-dwelling area 2-3 feet above grade. The hottest it ever gets down there in the summer is about 70F so I suppose it probably stays close to 55F or 60F through our winter (though I’ve never characterized).

Before I get to carried away with a heat pad, etc, I’ll confirm how my cells perform at 50F - unless there is a significant drop-off in capacity (as I suspect), I’ll skip the heat pad and just focus on insulating the cells to make use of the heat generated through the charge/discharge cycle...

Like I said, I haven't been counting the heat generated by the cells, and now you are pointing out heat generated by the buss bars. I'll have to noodle on that a bit. You are using 4 mOhm, and I would guess good bus bars are way less than that. I guess I need to go do some more research!

My estimates were 0.25 mOhm of internal resistance + 0.05 mOhm of resistance through each busbar + another 0.2 mOhms of contact resistance per cell (0.1 mOhms per Busbar-terminal contact).

0.45 mOhms / cell x 8 cells = 4 mOhms.

The easiest way to measure the resistance that matters once you have your battery is to use the R=dV/dI method:

Measure first voltage (V1) at a first lower current such as 10A or 20A (I1)

Now crank up current to a higher level like 50A or 80A (I2) and measure second voltage V2.

deltaV = V1 - V2
deltaI = I2 - I1

Rbat = deltaV / deltaI

All instantaneous lost voltage when current increases is from IR losses through cells, busbars, and connections, and all of that IR loss is translated to I^2R Watts of heat generated...

(obviously, over time, increased current also results in voltage drop due to cells being discharged, which is why it is important to get an immediate / instantaneously voltage measurement as soon as current has increased...).

 

[Horsefly]

I got my definition:

Rival in (K⋅m2/W) is the R-value

from here: https://en.m.wikipedia.org/wiki/R-value_(insulation)


If that’s not the correct R Val definition to use with the foam insulation we purchase at Home Depot or for soft wood, I appreciate an education.

The R value used in the U.S. is in °F*ft²*hours/BTU. Typical of us Americans, it isn't used anywhere outside of the U.S. (I think). Outside of the U.S. they use RSI (R-value Systeme International), which has the units in the wikipedia article: °C*m²/W. To convert from R to RSI, you divide by 5.678. Here's an article about the difference, and it's from SFGate, your hometown source! https://homeguides.sfgate.com/rsi-insulating-70075.html

 

[Horsefly]

My estimates were 0.25 mOhm of internal resistance + 0.05 mOhm of resistance through each busbar + another 0.2 mOhms of contact resistance per cell (0.1 mOhms per Busbar-terminal contact).

0.45 mOhms / cell x 8 cells = 4 mOhms.

The easiest way to measure the resistance that matters once you have your battery is to use the R=dV/dI method:

Measure first voltage (V1) at a first lower current such as 10A or 20A (I1)

Now crank up current to a higher level like 50A or 80A (I2) and measure second voltage V2.

deltaV = V1 - V2
deltaI = I2 - I1

Rbat = deltaV / deltaI

All instantaneous lost voltage when current increases is from IR losses through cells, busbars, and connections, and all of that IR loss is translated to I^2R Watts of heat generated...

Ok, I get how you got there. I don't have the big LFP cells yet, so I can't do the real world dV/dI, so I'll take your word for it. Sounds like it may be worth considering the heat generated by the pack, even in my relatively low average current situation.

 

[fafrd]

The R value used in the U.S. is in °F*ft²*hours/BTU. Typical of us Americans, it isn't used anywhere outside of the U.S. (I think). Outside of the U.S. they use RSI (R-value Systeme International), which has the units in the wikipedia article: °C*m²/W. To convert from R to RSI, you divide by 5.678. Here's an article about the difference, and it's from SFGate, your hometown source! https://homeguides.sfgate.com/rsi-insulating-70075.html

Thanks. Didn’t know that. I’ll adjust my calculations accordingly (and you might think about contacting Wikipedia since I didn’t see any mention there [though probably didn’t read carefully]).

 

[fafrd]

Ok, I get how you got there. I don't have the big LFP cells yet, so I can't do the real world dV/dI, so I'll take your word for it. Sounds like it may be worth considering the heat generated by the pack, even in my relatively low average current situation.

The main thing you want to avoid if you can is any need to also cool your cells.

 

[Bill Taylor]

OK, so I got the first round of testing done. This is a quick update, when I get the data collected I will present it in a more organized fashion. I used a chest freezer at 0°F. I repeated the test using three different 100ah non-heated lifepo4 batteries, to get an average. The results were nearly identical, so for the sake of time I will just use one going forward. For consistency, all batteries were charged completely at room temperature @ .2C and placed in the freezer for 12 hours before the discharge test began. I have a few preliminary thoughts:
1. The batteries warmed up *much* faster than I thought they would at .25C discharge.
2. I assumed the charge inhibit function would allow a very small charge in, even at freezing temperatures, this was not the case. I had to move the battery into the refrigerator overnight and warm it up to 38°F. I charged the battery at .25C, it warmed up much less than it did when it was discharged in the freezer.
3. I will use my IR camera next time to determine if the heat is coming from the cells or the BMS, and attach those images to the final report.
4. I also tried identical testing with a brand new AGM battery. I was expecting a severe reduction in power... not true. It did produce less, but considering it was at 0°F, it held up surprisingly well.
5. This is a lot more time-consuming than I originally thought. I do not have any data logging equipment, I'm doing it the old-fashioned way with a notebook and pen. It will take a few weeks or more to complete the testing. The internet is full of theory and assumptions. I'm looking forward to getting some real-world results together to serve as a reference. Stay tuned.

 

[curiouscarbon]

still reading the whole thread but thought this might fit for support https://www.lairdthermal.com/thermal-wizard/thermal-wizard-enclosure-calculator

 

[Horsefly]

1. The batteries warmed up *much* faster than I thought they would at .25C discharge.

I guess I would have thought the same as you: That a bigger current would be required to actually warm the cells.

2. I assumed the charge inhibit function would allow a very small charge in, even at freezing temperatures, this was not the case. I had to move the battery into the refrigerator overnight and warm it up to 38°F. I charged the battery at .25C, it warmed up much less than it did when it was discharged in the freezer.

That's a tribute to your BMS and it's low-temp cut-off. What BMS do you use? Does it allow you to set the low temp?

5. This is a lot more time-consuming than I originally thought. I do not have any data logging equipment, I'm doing it the old-fashioned way with a notebook and pen. It will take a few weeks or more to complete the testing. The internet is full of theory and assumptions. I'm looking forward to getting some real-world results together to serve as a reference. Stay tuned.

All I'm working with is "theory and assumptions" so you are doing us a service. Thank you @Bill Taylor !

 

[Horsefly]

still reading the whole thread but thought this might fit for support https://www.lairdthermal.com/thermal-wizard/thermal-wizard-enclosure-calculator

That looks pretty interesting. I plugged in my numbers, except for the thermal conductivity, which I would think is just the inverse of the R-value, thermal resistance. But that doesn't seem to work, as it makes the loss in my hypothetical numbers I posted earlier much higher than I calculated. I need to look through the info on the site. Thank you @curiouscarbon for posting it.

 

[Bill Taylor]

Update #2: I find myself repeating some tests several times. The results I am getting are in contradiction to my previously held assumptions and beliefs. I am using the same batteries, the same charging conditions, the same discharge machines. And I get the same results every time. This forum, and the internet in general, is full of speculation about how batteries work. There is a serious lack of unbiased information based on real-world test results. I am beginning to believe that almost everyone misunderstands how batteries work and just repeats what everybody else says. For example, it's common knowledge lead acid batteries lose most of their power when they get cold, right? Especially below 0°F. Not true. I tested three different AGM batteries, three times each. The average capacity loss at 0°F, compared to 70%, was 27%. That is a lot better than I expected. Also, Lifepo4 batteries warmed up quite nicely under moderate load. I tried four different 100ah batteries, their rise ranged from 8° to 16°F in just 15 minutes. This was with no enclosure or insulation of any kind, in a freezer at 0°F the entire time. I can't find where anyone has ever tested this before. I wonder if it's really necessary for people to build insulated boxes and heaters, when they could just have a small load on for a few minutes to warm the battery up enough to take a charge. But that all depends on how cold the battery is to start with, of course.
I thought this journey would take a few days. I can see now it's going to take quite a while. I'm continuing to collect information, I will post results as they come in.

 

[Horsefly]

For example, it's common knowledge lead acid batteries lose most of their power when they get cold, right? Especially below 0°F. Not true. I tested three different AGM batteries, three times each. The average capacity loss at 0°F, compared to 70%, was 27%.

I think that is pretty close to what I've read elsewhere, like at Battery University. Lead-acid batteries temporarily lose capacity when it is really cold, and regain it when it gets back up to 25°C. High temps cause permanent loss. I did some searching just now, and it seems most of the data I could find is about flooded batteries, rather than AGMs. However, I think the cold weather effects should be pretty similar.

That is a lot better than I expected. Also, Lifepo4 batteries warmed up quite nicely under moderate load. I tried four different 100ah batteries, their rise ranged from 8° to 16°F in just 15 minutes. This was with no enclosure or insulation of any kind, in a freezer at 0°F the entire time. I can't find where anyone has ever tested this before. I wonder if it's really necessary for people to build insulated boxes and heaters, when they could just have a small load on for a few minutes to warm the battery up enough to take a charge. But that all depends on how cold the battery is to start with, of course.

That is pretty fascinating. When you say a "moderate load" - Do you know what C rate you were using? I've been doing some discharge testing in my unheated garage. My focus was not on temperature changes, but I have felt the cells and could not tell any change. Next time I do it I will get out my digital IR gun and check for sure.

When you've seen the temp of the cells go from 8°F to 16°F, were they still in the freezer? If the freezer was at 0°F and the cells were 8°F, that probably means that the surface of the cells were 8°F, and the internal volume of the cells was probably warmer. Obviously there's no way to tell for sure.

I think the use case for many people may fit with your idea of just a periodic "moderate load" may be enough. However, depending on just how cold it is and just how often the cells need to be warmed, it may not work.

For my cabin, I would expect the ambient temps could be well below zero (I've got a temp logger up there for the winter, so I should know for sure once we can get up there). So my issue is how we handle an LFP battery when the cabin is unoccupied for 5 months or so. If we have to carry them out for the winter, that could be a deal-breaker, and I may be stuck with AGM. In this case I'm not sure a small load periodically is an option.

Great stuff though @Bill Taylor and I hope you can give us some C rates.

 

[Bill Taylor]

One of the batteries went up 8°F, another one went up to 16°F. All of them were kept in the freezer the entire time discharge rate was .25C (25 amps).

 

[Horsefly]

One of the batteries went up 8°F, another one went up to 16°F. All of them were kept in the freezer the entire time discharge rate was .25C (25 amps).

Ahhh. I read your post too quickly. So it sounds like you are pretty confident the cells were actually at 0°F when you started. Thanks for clarifying.

 

[Bill Taylor]

Yes, temp sensor inside batt case...

 

[diyernh]

Bill,
Thanks a lot for your work. I'm also looking to "store" LiFePO4 batteries in a cold camper in the Adirondacks. My solar charge is sporadic with the snow, clouds, and generally the low solar during the winter. I was considering using the 24VDC direct solar feed (before the controller) to dump the solar heat to the batteries. IF they warmed up, the heater would shut down, feeding the full wattage of the panels to the controller and charging the batteries.

Was your internal temp sensor in between the cells in the pack, or possibly near the BMS or ambient inside air? Did you cover the sensor to insulate it from air to only read the cell surface temps?

I was concerned about using the battery's power to heat itself. If the sun was only available for a short time, then gone, it could deplete the battery by heating itself with no charging obtained from solar. Figuring .25C discharge on a 100AH battery, I could get 4 hours before the battery dies and shuts down the BMS. If I can heat the battery in 30 minutes, I get about 8 days(30 minute warmups) with no solar charge. Once the BMS shuts down due to low voltage, there could be issues with the solar controller due to the warnings (Never disconnect the battery without disconnecting the solar first).

I'd be interested in using the battery charge to heat itself with say a 30W heating pad on the bottom. I think your current 25A, 12V or 300W would be too hot for this test. .025C discharge of 30W would heat from the bottom, with some internal heating of the cells.

Is it possible to test this to get the temperature rise for different time periods? Given the dT curve, others could extrapolate startup at any temperature. Other guestimates will be necessary as your test setup would be uninsulated, any additional insulation would preserve the heat inside.

My 800W of solar panels will peak out at 80W ( for only one hour) in the winter. That's not everyday, but one out of 4 days. Usually the clear days run together, followed by days of clouds/snow. If I can self heat the battery with it's own power at .025C discharge (30W), I could theoretically switch the heater on with solar voltage, using the battery to heat itself quicker. .025C on the 100Ah would be 80 days at 30 minute warmups.

I'm new to LifePO4, and just bought one Big Battery 170Ah. I'm hoping to work a heating scheme without opening it for testing and voiding the warranty.

Thanks
Carl

 

[Bill Taylor]

The testing continues.... I would like to do at least 6 or 8 charge-discharge cycles with each battery to average out anomalies and get good numbers. I do not have data logging equipment, I am keeping everything handwritten on a clipboard. I'm looking for an easy way to plot that information on a graph, I may just draw it by hand if I have to.
The temperature sensor was placed on top of the cells covered in heat sink compound, below the BMS.
Good idea, I will do a few tests @ .1C and .05C, and a few with each battery wrapped in fiberglass insulation inside a styrofoam cooler, just to get a comparison.
I grossly underestimated the amount of time and effort involved in this venture. But due to the glaring absence of reliable information about this topic online, I feel compelled to see it through. I have learned more about lithium cells, BMS boards, and charging techniques than I thought I would. I can't wait to get everything together, so I can sit back and watch everyone argue and nitpick over it. (Only slightly kidding)

 

[diyernh]

The testing continues.... I would like to do at least 6 or 8 charge-discharge cycles with each battery to average out anomalies and get good numbers. I do not have data logging equipment, I am keeping everything handwritten on a clipboard. I'm looking for an easy way to plot that information on a graph, I may just draw it by hand if I have to.
The temperature sensor was placed on top of the cells covered in heat sink compound, below the BMS.
Good idea, I will do a few tests @ .1C and .05C, and a few with each battery wrapped in fiberglass insulation inside a styrofoam cooler, just to get a comparison.
I grossly underestimated the amount of time and effort involved in this venture. But due to the glaring absence of reliable information about this topic online, I feel compelled to see it through. I have learned more about lithium cells, BMS boards, and charging techniques than I thought I would. I can't wait to get everything together, so I can sit back and watch everyone argue and nitpick over it. (Only slightly kidding)

Bill,

Maybe a video camera watching a temp sensor and a clock? Speed it up and put it on the web. Let anyone watch/record their own data points. I don't expect a perfect curve drawn. You could probably write down the temperature and time for the first maybe 30 to 45 minutes, then maybe come back to it when it's stopped moving. The curve could be drawn by hand.

I don't know where the BMS is located in your battery. Do you think it's heating your sensor, or you are further away from it.

Each battery is different. The BigBattery is a metal case with the cells laying horizonal. The BMS is in a different metal compartment on the top. Will has shown that they don't touch the cells with their temp sensor. This could end up with quite different temperatures between the air and the real cells.

I have my Big Battery outside at 16F(-9C) now. I just started the generator and it didn't put any charge into the battery. That's a good sign, or I don't have it hooked up correctly.... I'll try it over the the next week to see when the BMS lets it start charging. It's still a complex thermal issue as the batteries are in a different metal section. I could heat the batteries, and never warm the temp sensor, or visa versa. Probably the best for me is to insulate the entire battery with a heater near the cells. I could use your estimated time period as a delay to soak the entire battery at temp for xxx minutes.

I just noticed my Victron Solar Charger uses the battery temp sensor to inhibit cold charging. The Big Battery uses the quick disconnect plugs, There is no easy way for me to put the temp sensor on the battery post, directly reading cell temps. I may have to change the charger inhibit temp, or create a thermal mass mockup to put inside the battery box insulation. This issue is probably the same for all LiFePo4 batteries. Any temp sensor you put on the battery terminal is nowhere near the cells. It's just a small terminal connected with wires through the BMS. It could be way off from the cell temps.

You can probably work out a few rules of thumb, ie "The Taylor Principles". A 30 lb 100 AH Lithium battery in a 1" foam box, will reach cell core temp of 10 degrees above ambient with a .05C discharge in 20 minutes. 20 degrees in 30 minutes, 30 degrees in 50... Just discharge the battery into a heater at that load for the proper time and you should be fine. No one will survey their own batteries to prove their thermal issues. Manufacturers will not care that the terminal temperature, or the BMS sensor isn't reading correctly. They get to sell another battery if you damage this one with cold charging.

Carl

 

[Bill Taylor]

I have a battery in the freezer now. I will start the .1 C test tomorrow. The temp sensor is maybe an inch below the BMS. I'm taking IR images with a FLIR, it doesn't look like the BMS is heating up as much as I thought it would. I will put a few of those pictures up. After I do a few of the lower rate discharge tests, I will duplicate them with the batteries insulated. It should make a nice addition..
I have been working with Lifepo4's and using them in my shop and RV for years. But since I started doing this testing, I've been writing everything down, taking notes and really paying attention. I have exchanged phone calls and emails with business and industry contacts, and learned more in the last few weeks than I have in the last 5 years.
I'm also finding out a lot about charging those batteries. A lot of my misconceptions about proper charging are being revised, as I am doing as much charging as I am discharging.
I'm looking forward to condensing everything down to a simple presentation. It's been enlightening so far

 

[schroederjd]

I grossly underestimated the amount of time and effort involved in this venture. But due to the glaring absence of reliable information about this topic online, I feel compelled to see it through. I have learned more about lithium cells, BMS boards, and charging techniques than I thought I would. I can't wait to get everything together, so I can sit back and watch everyone argue and nitpick over it. (Only slightly kidding)

Hi Bill, First, thanks so much for contributing all the excellent content to this thread. I'm really enjoying the process of learning about how these cells work, and all the systems that go into keeping them happy. I wanted to address your comment about the amount of time and effort involved in the studies you're conducting. Just a couple months ago, I purchased my first Arduino (a MKR1000), and I'm constantly amazed at how much you can do with them. For a project like yours, with a $5 microcontroller, and a couple bucks in components (NTC thermistors, wires, etc.), you could be monitoring and logging temps, voltages, and other attributes real time. I had never done any coding before getting an arduino, and it does take a bit of self-study, but it's pretty straight forward. I'd be happy to help you with the setup if you're interested.

As an example of what you can do with a pretty simple set-up, here's some data I logged last night during a study of my cell warming setup. This is basically a test performed at 20watts for the first 12 hrs, followed by 5 watts of heat for the remainder of the test. I'm heating a 48V bank of 272Ah cells. A bit more detail about the setup I'm developing can be found here. I'll try to post some more info on these test results once I've had a chance to play with the data.

 

[Horsefly]

Hi Bill, First, thanks so much for contributing all the excellent content to this thread. I'm really enjoying the process of learning about how these cells work, and all the systems that go into keeping them happy. I wanted to address your comment about the amount of time and effort involved in the studies you're conducting. Just a couple months ago, I purchased my first Arduino (a MKR1000), and I'm constantly amazed at how much you can do with them. For a project like yours, with a $5 microcontroller, and a couple bucks in components (NTC thermistors, wires, etc.), you could be monitoring and logging temps, voltages, and other attributes real time. I had never done any coding before getting an arduino, and it does take a bit of self-study, but it's pretty straight forward. I'd be happy to help you with the setup if you're interested.

As an example of what you can do with a pretty simple set-up, here's some data I logged last night during a study of my cell warming setup. This is basically a test performed at 20watts for the first 12 hrs, followed by 5 watts of heat for the remainder of the test. I'm heating a 48V bank of 272Ah cells. A bit more detail about the setup I'm developing can be found here. I'll try to post some more info on these test results once I've had a chance to play with the data.

View attachment 37803

First off - I agree in thanking @Bill Taylor for the hard work he is putting in to contribute to the greater knowledge about how these cells warm during discharge. I didn't account for that at all in the computations I supplied at the beginning of this thread.

Re: Arduino for logging, I am considering putting something together myself. I was thinking about using an ESP32s or ESP8266, and coding via the Arduino IDE. Did you add an SD card to your Arduino to store the logs? I have a breadboard ESP8266 collecting outside temps and sending them via HTTP Wi-Fi to a server I have in the house. I could probably use that same design (and most of the same code) to log the temps of LFP cells.

I've also been thinking about doing another microcontroller up at our cabin to log the Ah going in and out of the battery bank over time. One of the issues I have in considering a migration from AGM to LFP is that I don't think we really know what our energy consumption is over time. My wife and I go up there way less than either my brother's family or my sister's family, so it's hard to know how the system is really doing with no one there observing it. I'd need to think through reading the voltage divider data from a shunt, and I'd need to buy a SD card add-on, which should be easy.