Well they liquid cool some gaming computers, so why not? I can just see it now, rack batteries, not just with active balancing, liquid active cooling and for good measure throw in active compression. I’m sorry I was on a roll?
What’s your daily SOC range?
- Thread starter Skypower
- Start date
High heat reduces the life of lithium. The cost of immersion cooling isn't high. Temperature gradient across cells reduces the life. All the heat comes from the +ve cell terminal when charging/discharging. The temperature gradient across the cell reduces life expectancy. Add high ambient temperature on top of this aging effect and you get greater reduced life. An aquarium chiller works well. You don't need to spend lots of money cooling a room. You just insulate the battery and cool it directly. Keeping the cells at a constant 30degC in summer would be efficient for the cooling system and life expectancy.
“An aquarium chiller works well.” Intriguing, never thought of that. You may find out that synthetic oils are poor carriers of thermal energy however are very safe in contact with most aluminums(cell case). Distilled water is one of the best carriers of thermal energy but it’s somewhat corrosive to aluminum even with glycol. I wouldn’t want it in direct contact with the thin skin of a cell. Some sort of cooling jacket using water OR loose the refrigerant to water heat exchanger for air type and blow air on the cells. Just can’t find tiny AC units as small as these chillers for cheap.
robby
Photon Vampire
- Joined
- May 1, 2021
- Messages
- 4,296
I agree with the overall conclusion of these charts but question the figures for capacity loss.
I am wondering how the battery manufactures can Warranty batteries to have only 20% capacity loss over 10 years when this data is showing that even under ideal conditions there would be a 30% loss in 10 years. According to those numbers a typical pack at lets say 30degC would only last about 2 years before it lost 20% of it's capacity.
Am i missing something about this in the chart data?
I understand that cycling will lessen the loss if done correctly, but to what extent?
I’ve asked myself the same question. Keep in mind that there have been subtle changes to the design and chemistry in the last few years. So if ten years of present design hasn’t occurred yet????? Theoretical projections based on partial degradation?
Couple of things:
- The tests in the paper are done with small cells, that likely impacts the data
- The tests in the paper did not identify if the capacity loss measured has the same trend after the initial year test. Normally, the capacity loss will get less year over year. It might start at 3%, but it won't be 3% the year after that.
- Manufacturers typically qualify a cell based on calendar age, but rather cycle life (at 25C). If you cycle the cell 4000 times in the first year, you get to 80% of its original capacity. That does not mean the cell will live 4000 years if you cycle it once a year.
- Calendar aging is a thing no matter what state of charge or temperature (and the numbers in the paper don't consider these separate), although low temperatures can slow this down significantly. You could store the cells at -40C to keep them well below the degradation numbers in the paper.
We are way down the development phase. The oil has been used extensively for cooling of HV transformers. We have a battery on test. It's been sitting in the oil for many months. Winston cells. Ditto a BMS. All working. Components are compatibility tested first before immersion.
The EV car industry are developing this immersion cooling using the same fluid in direct contact with cells
littleharbor2
Solar Addict
True, very inefficient. I have extra batteries and panels laying around so built my own dedicated active cooling system. In this heat lately it still struggles to stay below 85 f. In this humidity it creates a lot of frost/condensation so when I built it I formed a chute for the condensation to run out of the box on.
Attachments
Re: SOC of LiFePO4 batteries, I just discovered the value of a shunt. Here's the story.
I installed my small (1.2KW) 24V system to power an off grid shed we built for an exercise studio, along with the lights in our wood-fired sauna. I started with a Victron Multiplus and a MPPT SCC connected to an RPi running their Venus software. One Amperetime 100 ah LiFePO4 battery to start, then added a second one about a month later, just as winter started last year. Heat in the studio is from a propane Rinnai wall heater, which requires a small amount of electricity. I was just using the voltage-based measurements from this system to estimate SOC. Even with our short days in winter, I only had to fire up a generator 2-3 times to keep the batteries charged above 20-30%, which is as low as I wanted to go, "just in case" (I know I can safely go lower, but I always like to have a reserve, for everything).
When summer arrived, I decided I liked the location of the studio so much I decided to move my office up to the studio, and build another building for the studio (that'll be happening now through fall). But I needed a lot more AC in that building to stay comfortable, so the voltage/system ESTIMATED SOC could fall as low as 30-40% even on sunny days, and the panels are shaded in the afternoon. I didn't want to worry about that when it's cloudy and rainy but I still need AC to combat humidity. So, I ordered an additional (third) battery to ensure I had it before the looming UPS strike hits, and at the same time decided to finally break down and order a Victron SmartShunt.
As it turns out, I should have installed that SmartShunt on day one. The non-shunt estimates from the system were clearly underestimating the battery SOC, and I probably had no need for that additional battery (shhhh... don't tell the Mrs). My ACTUAL SOC is rarely dropping to even 90% on days with decent sun, and I'm betting it won't drop to even 65% with 2-3 days of rain and clouds (about to find out). Anyway, the moral of the story is that a good shunt may end up saving you money. I'm not unhappy that I have the extra battery capacity as it will give me more peace of mind if we're traveling in the winter, but had I installed the shunt first I might have one less ~$700 battery. Feel free to learn from my mistake.
I installed my small (1.2KW) 24V system to power an off grid shed we built for an exercise studio, along with the lights in our wood-fired sauna. I started with a Victron Multiplus and a MPPT SCC connected to an RPi running their Venus software. One Amperetime 100 ah LiFePO4 battery to start, then added a second one about a month later, just as winter started last year. Heat in the studio is from a propane Rinnai wall heater, which requires a small amount of electricity. I was just using the voltage-based measurements from this system to estimate SOC. Even with our short days in winter, I only had to fire up a generator 2-3 times to keep the batteries charged above 20-30%, which is as low as I wanted to go, "just in case" (I know I can safely go lower, but I always like to have a reserve, for everything).
When summer arrived, I decided I liked the location of the studio so much I decided to move my office up to the studio, and build another building for the studio (that'll be happening now through fall). But I needed a lot more AC in that building to stay comfortable, so the voltage/system ESTIMATED SOC could fall as low as 30-40% even on sunny days, and the panels are shaded in the afternoon. I didn't want to worry about that when it's cloudy and rainy but I still need AC to combat humidity. So, I ordered an additional (third) battery to ensure I had it before the looming UPS strike hits, and at the same time decided to finally break down and order a Victron SmartShunt.
As it turns out, I should have installed that SmartShunt on day one. The non-shunt estimates from the system were clearly underestimating the battery SOC, and I probably had no need for that additional battery (shhhh... don't tell the Mrs). My ACTUAL SOC is rarely dropping to even 90% on days with decent sun, and I'm betting it won't drop to even 65% with 2-3 days of rain and clouds (about to find out). Anyway, the moral of the story is that a good shunt may end up saving you money. I'm not unhappy that I have the extra battery capacity as it will give me more peace of mind if we're traveling in the winter, but had I installed the shunt first I might have one less ~$700 battery. Feel free to learn from my mistake.
Last edited:
littleharbor2
Solar Addict
Shunt based meters are a nice addition but doesn't your BMS give you accurate SOC info. They are really just a glorified shunt programmed for your battery.
timselectric
If I can do it, you can do it.
- Joined
- Feb 5, 2022
- Messages
- 19,796
BMS SOC readings are notoriously inaccurate.
littleharbor2
Solar Addict
I didn't realize that. They show current in and out and have voltage numbers down to the thousands of a volt. I'm surprised they would be inaccurate.
timselectric
If I can do it, you can do it.
- Joined
- Feb 5, 2022
- Messages
- 19,796
I was also surprised, at first.
But a smart shunt is the most accurate thing that I could find.
Especially if running multiple batteries.
One true SOC for the entire bank.
EastTexCowboy
Solar Wizard
Having done both I concur. I don't think it hurts to occasionally look at individual battery state but for day to day this is by far the best. Of course I only went that way because of advice from you and a few others here, so thanks. ?
The difference between accuracy and precision.
littleharbor2
Solar Addict
Yes, I am surprised. Good to know because I have a "dumb" BMS on my main battery and a Bogart tri-Metric giving me Soc.
Simply stating that water is by far the best heat carrying fluid, tho it has drawbacks which is why there’s transformer oil. To compensate the different fluid through the heat exchanger must move more volume or redesigned for the medium to achieve the efficiency rating of the chiller IF that’s even necessary.
W/m K;
Ammonia .507 nasty
Ethylene glycol.258 toxic
Transformer oil .136 low reactive
Water .609 mild reactive subject to freeze.
Winstons are excellent cells, very expensive(4X?) and unlike the vast majority of the cells used on this forum they use a plastic case.
I very much believe that your project works ,however consider costs to be marketable. If the system, cells+bms+chiller+housing is four times the cost, well I’ll just buy four batteries as needed.
Looks like I’m replacing my battery meter(Hall effect type) It’s been getting worse. It’s only reasonably accurate if it gets to a full state of charge within a couple days and when it gets hot the voltage is incorrect. It has lots of nice info, pretty colors and…it’s junk. Got another cheapie brand, “insert clone name here”(shunt type) that’s been laying around in a box for a year and tested. I’ll use what I’ve got and If it’s a fail I’ll just cave in and get a Smart shunt. The JK bms are reasonably accurate if the voltage and current it records is correct. And if you adjust it in the app you need to use a very accurate meter/Fluke. Edit; (AND a very steady current draw as reference!)
I have three JK’s and all were off except the last one had the correct voltage.
Last edited:
Quattrohead
Solar Wizard
Till they start sending SOS messages by smoke signals ?
timselectric
If I can do it, you can do it.
- Joined
- Feb 5, 2022
- Messages
- 19,796
Nah
They're very happy with the process. No overcharging, overheating, or over working.
robby
Photon Vampire
- Joined
- May 1, 2021
- Messages
- 4,296
They only drift out of accuracy if they are not calibrated often. So long as you get a 100% charge at least every 48 hours they remain accurate. More expensive BMS systems can be even more accurate.
I use to have both in my system but removed the shunt because there was no significant difference in the reading.
My personal opinion is Solar Assistant should put high priority on integrating Smart shunt with data from popular BMS’s (if that’s even possible). Right now it’s one or the other. Alternatively, perhaps disable one or the other through the app? I like simplicity and seamless functionality.
Last edited:
timselectric
If I can do it, you can do it.
- Joined
- Feb 5, 2022
- Messages
- 19,796
Mine are calibrated daily. And drift daily. Two will report 100% while still drawing 20a charge for a few more hours. Three of them only reset to 100% if they see an over voltage alarm. Don't get me wrong, they are usually within 10% of the true SOC. But I don't call that accurate.
Similar threads
