Cell balancing @ partial charge?

[Tonni Lerche Sørensen]

Hi !

I am looking into using LiFePO4-batteries as standby batteries - and trying to figure how extend calendar life by cell balancing.

So - LiFePO4 needs cell balancing - this normally happens when the battery has been charged up fully, where the voltage of each cell begins to variate alot.

But if we want to prolong calendar life, we can deliberately buy a too large battery and let it sit at partial SOC - this extends life of the cell (according to 3-4 scientific research projects i've found during my bachelor-project)

BUT - how about cell balancing at partial charge? - is it necessary? - when switching from constant-current to constant-voltage at say 50% SOC for a LiFePO4-battery, will the voltage of each cell still begin to differentiate? - will the BMS board still do a cell balancing when a charge voltage is not high enough to reach 100% SOC ? - and in what relation to active balancing or passive balancing?

Example of balancing:


(source: Eaton white paper regarding lithium in UPS application)


Best regards :D

Tonni Lerche Sørensen
Marine engineer (to-be @ june)
Denmark

 

[MisterSandals]

will the BMS board still do a cell balancing when a charge voltage is not high enough to reach 100% SOC ? - and in what relation to active balancing or passive balancing?

I think BMS's only balance when in very low SOC and very high SOC. And even then, only when the out of balance condition meets a certain threshold.

@BiduleOhm is creating a BMS and is probably a great source of info on the low down specifics of what you are asking.

From what I've read, most BMS's are different in their balance start thresholds, balancing methods and the balancing current. And the difficulty of uncovering all this info is part of what makes it so difficult to choose a BMS for your particular use.

I know thats not much help, sorry. But hopefully BiduleOhm reacts, he will get a notification from the incantation above.

 

[BiduleOhm]

The short and quick answer is: it depends on the BMS.

Now, to elaborate a bit more because I guess the short answer isn't really useful ?, most BMS (all?) do balancing only at high and/or low state of charge because the physics of Li batteries make the voltage curve really flat inbetween and the BMS use voltage to compare the cells and decide which one is the most charged one, etc...

Now @MisterSandals actually suggested to me a very good idea which is to use relative capacities and SoC instead of voltage to decide which cell is the most charged, etc. (at least for the flat part of the curve, you can still use classic balancing technics for the top and bottom ends). By doing that you don't care about how flat the curve is as you're not dependent on the voltage anymore. The main problem is accuracy as SoC is time related so it implies derivation and integration which implies cumulative error terms (for example anyone who tried to derive a position from an accelerometer knows what I mean...). It can get really complicated really quick and it's not really the topic here but I had to mention it as it's a big thing to think about when using the SoC method. It also needs some cycles to learn the relative capacity of each cell (unless you measure and input them manually in the BMS when you build the pack of course).

As far as know no BMS on the market uses this method yet. I don't know if they tried but then found it was a bad idea, or if it's just too complex and not cost effective, or if nobody thought of that before. I'll definitely try it on my BMS as I don't really care about cost effectiveness as I'm not trying to design a product to sell it, and as it's a very interesting idea.

Is it necessary? well if you have classic balancing at low and high SoC in addition then no, it's just nice to have. But in your case it might be the only balancing you can do (depending on a lot of things, mainly how much you oversize the battery and so stay only in the flat part of the curve) so then it's a yes if you want balancing. I say if as some people prefer to balance the pack manually once in a blue moon and let it live it's life the rest of the time (NB: I don't like this strategy as I think it's just an excuse to not have a proper balancing solution, but to each their own opinion of course).

Now a decent BMS will have user customizable parameters for things like start of balancing voltage, etc... so you can always set them lower than the usual values to start the balancing earlier on the charge curve. The problem you might have then is not enough accuracy on the voltage readings for the balancing to work as it should on a flatter part of the curve.

Active vs passive balancing isn't really relevant here has the only difference between the two is that passive balancing discharge the most charged cell into resistors and energy is wasted as heat while active balancing discharge the most charged cell and use this energy to charge the least charged one. In practice passive balacing is fine and active balancing is really for specific cases (and it's more complex so it's more expensive), Will made a very good video on that subject actually

Sorry in advance for any unclear parts or errors in the text, it's 3 am here...

 

[Tonni Lerche Sørensen]

@BiduleOhm - thank you alot ! - this is really helpful !

I see why using capacity and SOC to balance is hard to do, as you mention accuracy is the problem.

My plan was to let the battery stay at 30-50% SOC pretty much always - i've oversized to a factor 10, which means discharge between 30-40% or 40-50% - yes this means staying on the flat plane of the charging curve.

So if i were to buy a commercial 48v battery with unknown cell balance properties, i would only be able to expect balancing at top and bottom - so i would actually go for the blue-moon-strategy, because there is not much else to do? - meaning charging fully (from 30-50% to 90-100% (above flat plane) to let the cells balance - and then apply fake load to let the battery reduce to my desired "stored" SOC - and doing this once in a while, say for every three months or so?

The real discharge for the consumer will only be 1-2 times each year (48v backup power, when grid power is lost)

 

[Ampster]

Since the theme of this thread has become "it depends", I will add my two cents. It depends on whether the inverter has any overhead. My old inverter had several modes of backup. The fastest mode had a cutover time that was very fast, but it had 150 to 300 Watts of overhead. Another mode was slower to cut over and modems, PCs and microwave clocks had to restart or be manually reset. But that mode had very little overhead.
What I am suggesting is if your inverter has overhead and the inverter can be set to start charging at 50% you could let it cycle between your preferred charge finish point and charge restart. That way you would not have to devise an artificial load to get you to that sweet spot of long life.
Balancing is another subject. My initial belief is that with that conservative cycling your pack would not experience much drift. You could set the balancing to only occurred when deltas are significant or on some schedule where you would choose to take charging voltage to the point that your balancing would engage.

 

[BiduleOhm]

Ok, with that much of oversizing (that's crazy high BTW, initial cost will be through the roof) balancing becomes secondary as you'll never go even near the top/bottom SoC, so a good balancing at the start with some manual balancing from time to time will be fine

 

[MisterSandals]

Maybe a cheap balancer would work.
I have a $10 Aliexpress 4S balancer that does 1.2 amp balancing at all SOC.
It does not start to balance with less than .1v cell difference (far too course for my day to day use).

Might be good for large unattended battery bank? This comes in 16S, probably a bit more expensive.

(Sorry, cannot get link out of app)

 

[Tonni Lerche Sørensen]

What I am suggesting is if your inverter has overhead and the inverter can be set to start charging at 50% you could let it cycle between your preferred charge finish point and charge restart. That way you would not have to devise an artificial load to get you to that sweet spot of long life.

A charger with such properties is already in my plans - and i've found a few of those with ~ 50% overhead and able to finish at a programmed voltage, cut off after a certain time, restart again at a certain voltage level - and forced start after xx number of days - all programmed via software from a computer - the MEC-Energietechnic Nova-1000F and the Piktronic KOP400-48 - they are not the perfect choice for an industrial setting, since they only offer 1 year of warranty - for but building a proof-of-concep system they will suffice - i am also looking Mean Well products, they seem interesting though not as flexible in adjustments as the two former.

The idea of charging to 100% was ONLY to make cell balancing work :D

 

[Tonni Lerche Sørensen]

Balancing is another subject. My initial belief is that with that conservative cycling your pack would not experience much drift. You could set the balancing to only occurred when deltas are significant or on some schedule where you would choose to take charging voltage to the point that your balancing would engage.

So with say, 5-10 cycles each year, and a estimated calendar of 10 years @ 50% SOC (total ~ 100 partial discharges) - you dont think balancing will be relevant?

The battery i've been looking at has socalled "matched cells" - but i would expect that, since its a 48v 25Ah LiFePo4 with 26650-cells - and i believe it will only balance at each serial bank.

 

[Tonni Lerche Sørensen]

Ok, with that much of oversizing (that's crazy high BTW, initial cost will be through the roof) balancing becomes secondary as you'll never go even near the top/bottom SoC, so a good balancing at the start with some manual balancing from time to time will be fine

Ok here goes :D

Actually i only need 1-2 amphours @ 48 volts, discharged at 7-14A over eight minutes - but finding a 48V 3Ah commercial battery is impossible - DIY is not preferred since its for industrial usage (worries about warranty) - its 2 (or 4) x 48v DC motor using ~ 3,5A each over eight minutes (meaning recharge is possible after only one cycle).

Next best was a 48V 25Ah from a french manufacturer called PowerTech

Next solution (just arrived yesterday at my email inbox) is a special built 48V 10,8Ah battery from China with IFR26650 cells and BMS - which is 3 x cheaper than the Powertech - this size seems "okay" if I account for the capacity loss over 10 years, and thus letting it sit at 50% SOC.

------

So a manual balancing by charging fully and then reducing with a load - say once a year? - if i am able to monitor the cells from the BMS, I of course would be able to come up with a plan of such

 

[BiduleOhm]

Oh ok, it's for a really small battery, no problem with the cost then.

Yeah, once a year should be enough if the cells are matched and good quality

 

[Ampster]

So with say, 5-10 cycles each year, and a estimated calendar of 10 years @ 50% SOC (total ~ 100 partial discharges) - you dont think balancing will be relevant?

Sounds like you will have remote monitoring capability or at least some process to evaluate the system every six months. A cell that appears unbalanced could be a cell with a slightly lower capacity. That lower capacity only shows up near the top or the bottom so your strategy should mitigate any risk of that reducing the functionality of your pack.
There is a concurrent thread on top balacing where we got on a side track and discussed the advantages of lower cycle percentages. @nebster commented on his experience and research on cycle life and LFPs. That reinforced the thinking on this thread.

 

[Ampster]

Maybe a cheap balancer would work.

My guess, based on the critical industrial application he is using, it that he wants reliability and remote montoring and error reporting.

 

[Ampster]

There is another thread where a guy is using an industrial UPS for backup. Did you consider that?