LFP longevity vs SoC operating level

There's been a recent study (this year) run by the guy who patented NMC batteries - and gets paid by Tesla to do studies like this. Linkage - https://iopscience.iop.org/article/10.1149/1945-7111/ad6cbd

Basically the study looks at LFP batteries and attempts to determine the best SoC range to cycle the batteries in order to maintain lifespan.

There's a video relating to car batteries here - https://youtu.be/w1zKfIQUQ-s?si=tpeh69ptTbJ5OzkU - and if you skip to 10 minutes you'll see a graph showing the relevant ranges etc.

The tl;dr of the study is that the worst thing you can do in terms of maintaining battery lifespan is to regularly charge from 0-100%. ie the very thing which GE say you should be doing to "exercise" your battery. Edit - actually the absolute worst thing you can do is cycle between 75-100% SoC, which may be of interest to predbat users as I'm sure there's a way of preventing it.

The message is that running LFP batteries at low SoC (0-25%) is best practice and very occasionally (once a month) charge to 100% to help SoC tracking. We all know that if you do that on a GE battery then the SoC is complete nonsense.

No easy answers here for GE owners as your real-world choice is a SoC which is nonsense or a battery which will degrade faster.

Vestas which 0% and which 100% though?

The 0 (or 4) and 100 we see aren’t really 0/4 or 100. Except for forced calibrations.

TX200 100% on the app/inverter = 97% on the battery.

The lower end is actually correct. 4% on app/inverter = 4% on battery.

That of course assumes the SoC is correct. If its not then its quite possible to discharge to 0%, or very close to it (100Wh, 45V is the lowest I've seen) and of course its entirely possible to charge to 100% as that's when you see overvoltages. Or at least you did before GE hid the notifications.

There's no margins other than that on a 9.5kWh battery. In theory the battery should only charge to 186Ah, so when new its probably only charging to 94% actual capacity but that went out the window with v3015.

Same thing happened with the older batteries - ask @hoggy how hard his battery got hammered.

Anyway when they're referring to 0-100% in the study/video they're referring to the SoC as displayed to the user.

The GE app says my 2 batteries operate between 10% and 96% for the overall 4 and 100% points. But I don't have enough capacity not to use the full range so if I find my battery breaks within the 10 years (and GE are still around) I'm relying on the guarantee. But it will give me the opportunity to try other technologies (possibly V2H by then).

I think I'm just going to set the AC charge limit to 90% for a bit and see how far out of whack the SoC gets.

Makes sense for me as I'm on v3015 and that will cause the battery to be idle (cut-off limit) for a bit longer so should rebalance more regularly at the low end.

If it doesn't then it'll require external help from givTCP/HA.

Vestas

Isn’t there “fat” above the 0% and 100% …?

There must be, as during the calibration it goes much lower and much higher.

Mine typically give 0% at about 50v and 100% at about 53.5 or 54v.

During calibration they go down to 45v and up to 57v
Or so. So surely that’s the batteries ‘real’ 0-100% range. In which case they aren’t operated truly from 0-100% - more like something inbetween (say 10-90 or 20-80%) of their real capacity. We are just seeing our day to day operating range as an artificial 0-100% labeled by GivEnergy.

Pete UK yes that's what I thought

Pete UK i agree. I have a 13.5kWh AIO, but the raw sensor data indicates 16.48

Pete UK Not on the 9.5kWh battery. At the point of manufacture it has a 200Ah capacity so there's 14Ah "overprovision", or if you prefer 700Wh (0.7kWh) . That will more or less disappear in the first year or two due to SEI and other technical reasons. From that point onwards there's no overprovision.

The difference between 54.3V and 56V is only 3% of capacity. If SoC is accurate then the battery will charge to 97% while the inverter/app displays 100%. If SoC isn't correct then the battery will charge to 100% and that's when you get overvoltage notifications (or did before they were hidden).

As I mentioned above its entirely possible to go below 4% if the SoC is wrong. At the time of writing this GE haven't (yet) hidden undervoltage notifications and I have several of those where the battery went all the way down to 45V.

A recent calibration only took the battery to 46V and 56.6V so its entirely possible to exceed calibration limits if the SoC is wrong.

There is no "fat" 🙂

Isn’t it;

At 57v;
(200 Ah x 57v) ÷ 1000 = 11.4kW
At 56.6v;
(200 Ah x 56.6v) ÷ 1000 = 11.33kW
At 53.5v;
(200 Ah x 53.5v) ÷ 1000 = 10.70kW

Or even at 53.5v with 186Ah;
(186 Ah x 53.5v) ÷ 1000 = kW = 9.951kW

And after a few years the 0% and 100% window labeled up by GE would be shrinking to a less usable range as well?

I empty mine every night and fully charge them but I don’t see 10.7kW going in from the grid. During a calibration I see more going in though.

Just my observation.

Pete UK No because the operating voltage is 3.3V per cell and hence its 52.8V The charging voltage spike on LFP batteries is rapid and usually within 4% of maximum cell capacity. Its part of the charge graph and BMS firmware is set to look for one cell at 3.6V or two cells at 3.55V as part of calibration just for that reason.

You appear to be missing the point anyway - the ranges in the study/video refer to the SoC presented to the user.

All manufacturers have software which doesn't present the real SoC to the user. Pick a mobile phone, any phone 🙂

Reason is obvious - discharging to 0% = physically damaged cell = warranty claim. Ditto 100%.

What the study/video is saying (and its worth watching if you do have an EV) is that for LFP batteries then the second worst thing you can do is regularly cycle them between the maximum limits imposed by the manufacturer.

Except of course to maintain SoC accuracy you need to charge to 100% once in a while (week/month for some batteries) but for GE's current range of low-voltage batteries that means doing it probably twice a week.

Hence the dilemma.

It will be interesting to see what happens in a few years when GIvEnergy start performing the health checks they say they will do at years 5,8 and 10 for the 12 year warranty.

There is no mention of any limit to the number of battery-cycles within the 12-year period and those of us on the Cozy Octopus tariff have 3 daily opportunities to export large chunks of our capacity.

If my usable capacity falls too far to make my regular exporting financially viable, I can simply switch to only exporting excess solar and I will still have plenty of battery capacity to meet my household needs assuming that the cozy Octopus tariff remains approximately where it is (and the export tariff also does). However I suspect that in the next 18 months, the difference between the Cozy rate and the export rate will persuade me that I should only export excess solar anyway. This will be minimal because of my smallish array and high daily use.

Weasel Well there's also the fact that the battery stops* working at 70% of the original capacity which I suspect most people won't be expecting.

*datasheets for LFP cells say you have to stop the battery operating if it gets to 150% original resistance or 70% of original capacity. There are good reasons for this as LFP cells in that range are liable to go pop on max rate charge/discharge.

Thanks for the heads up. I reckon my batteries would need to drop only to about 90% capacity for me to switch to exporting only excess solar

Weasel Well don't go doing that.

The 5000 cycles or so GE warranties on batteries is usually measured to 80% original capacity at the LFP cell manufacturers, not 70% so there's leeway there.

Vestas

Ok then.

Vestas I’m not

Vestas I’ve been tracking mine and I don’t think it’ll make it anywhere near 5000 cycles before it hits 70%. I’ve only done around 600 cycles and already lost 9% of original capacity. I estimate that given a small plateau I would get around 3000-3500 cycles out of it. If that then stops working at 10 years because I do 300 cycles a year and gets to 70% then it could have been one of the worst purchases I’ve ever made and cost me a lot of money so I hope that doesn’t happen!!

Al E.Gator Pretty much the same figures here. I guess we'll see what we see.

Hmm.. food for thought for me then.

Weasel I shouldn’t worry about it too much, GE say use it as much as you want and have warrantied it. If there’s a load of us whose batteries shut down after 8 years there would be major uproar with GE and potentially a lawsuit. That’s my attitude anyway. Though I am very concerned at the rate of degradation, an NMC powerwall has significantly better deg and I thought going with a GE LFP battery would give a much longer lifespan. Really I want someone from GE to comment on the matter.

Windy Miller warranty isn’t worth the paper it’s written on to be honest. If it breaks at 8 years you’ll only get a tiny amount of money given the sliding scale of compensation, certainly not enough to make up for the fact that I’m sure most people buy a battery expecting it to last 15+ years

Voltage, charge rate, charge temperature (too hot or cold) pack balance, quality of cells….

Or the system being completely unable to understand the power it has so getting the capacity wrong..

kram I do agree, but at the same time it measures should be taken (as with power wall for example) for GE to be able to manage all those things themselves. Particularly the price of these systems and the payback calculations. Not to mention binning a battery after just a few thousand cycles is unbelievably unenvironmentally friendly

Al E.Gator

PW2 has a complex cooling/heating system, uses nmc cells and has a better ability to track SoC.

It’s much harder to control some things, particularly temperature in a battery without the equivalent feature.

Vestas I think of the battery as a lift. You have room at top and bottom to enable it to settle at the right level. The calibration on my latest battery went from 0% at 45.6v to 100% at 55.7v. According to the Rubikscube monitor app its full capacity is 197Ah. At 4% after leaving it to settle it is 50v. If it grid charges then it stops charging when the app says it’s at 100% (97% by checking the battery details, 194Ah). When this happens the voltage has been 57v, with 2 cells over 3.6v and the other 14 cells over 5.5v. It then settles back to about 53.5v. If it is charged from solar then it stops charging at 100% (100% by checking the battery details, 197Ah). This goes up to 56v then settles back to 54v.