Over-discharge of LiFePO4 cells

[CatNewBee]

If you need 400Ah / cycle - it does really not matter for what.

The price per Ah is almost linear in the LFP world as in the FLA world.
There is a battery for every demand and pocket.

I am not saying, everybody needs 1000Ah, There are people with larger banks out there too..
If I recall it right, the Winns have 1400Ah LFP on board, I saw cats with batteries from 700Ah Winston cells,
also boats and RV's with 400Ah down to 100Ah. They use the power often for very different purposes,
some even as a start battery built from 40Ah cells - what is nonsense in my opinion.

There was a L440-OW liveaboard catamaran with a 123BMS and 1000Ah cells in the Med for sale last year on yachtwold.

So it is the tip of the iceberg. There is much more LFP out there on the water then you may think, even in old cheep RV, where the battery is more expensive than the vehicle itself.

Don't judge the people by their gear, it is always about priorities and for some Lithium makes sense - even financially.

[evm1024]

Originally Posted by Maine Sail

SNIP

My questions are always:

* Why push into the vertical when it's not needed for capacity?

* What benefit is there by pushing into that area where there is really no quantifiable extra capacity?

* What would the outcome on cycle life be if you did not push into that knee every cycle, even with a "stop charge" protocol?

If you find a paper that answers any of these questions please let me know.

Regarding the original post, and over discharge, I can only mention that the difference between a lab, and the pin point accuracy they have, and the real world is on the order of comparing apples to centipedes. The lower knee becomes so vertical below about 2.9VPC that the drop will occur really quickly and damage can be done, if the pack is not protected with LVC.

Also the A123 LiFePO4 nanophosphate cells (now owned by Lithium Werks) and the K2 cells are some of the best and most robust 18650 cells I have tested here. I simply can't seem to kill them but, building a DIY pack from them without the proper tooling is just not very feasible.

There really is no effective capacity to go after below about 3.0VPC. I have actually raised all my testing here to 3.0V as we are talking a minutia of capacity left and not worth risking even with test cells.

The original paper is exploring the limits and using 100% SOC, 100% DOD points that are the most (IMO) scientifically valid. I am sure that the authors and we who are reading this thread do not see those points as targets for our "house" banks.

Agreed that there is no need for any of our systems to go into the knees. The time from almost anywhere in the knee to over-discharge and damage is measures in minutes for a 400 AH bank discharging at 20 amps rate. Whereas the time from 80% DOD (20% SOC or 80 AH residual capacity) to 100% DOD on that same bank is about 4 hours.

The real take away from the paper was that a) going to 100% DOD does not destroy your bank (assumed to be 2.0 VPC) right then and there. and that b) hard cycling (at lease a A123 cell) from 100% SOC to 100% DOD still gives cycle counts in the thousands.



I do have an email into Dr. Xie asking if they would consider running a study where the cells are cycled between 80% SOC to 20% SOC so as to predict total cycle count. No answer yet. They may have such a study already that they have not published. Given that some of their funding comes from DOE and DOD sources it would not astound me that they do.

[foufou]

Why is the cost of batteries measured in $/amp hours rather than $/amp hours/cycle?

[john61ct]

Cycle data is grossly inaccurate, cannot be used to reliable compare quality from one maker to the next, only between lines within one brand.

Influence of marketing dept vs maybe-more honest engineers

Different lab protocols

And lab results do not come close to IRL data, and the latter can't be quantified, how the bank is cared for is at least as influential as initial quality when purchased.

$ / AH / year is what I use, but intuitive / WAG assumptions involved.

And of course weekend usage varies from full-timing.

[CatNewBee]

Originally Posted by foufou

Why is the cost of batteries measured in $/amp hours rather than $/amp hours/cycle?

because most people have no clue what a cycle is and it is a powerful marketing tool.
Even the industry does not have one.

To be most accurate, it should be $/kAh delivered during battery life, and also a clear point, when battery life is considered over (e.g. decline to 60% of nominal capacity).

Some count cycle as discharge to a value and re-charge.

So in this view a cycle to 20% DOD of a 100Ah battery is 80Ah and it can deliver lets say 300 cycles totaling 300x80Ah = 24kAh as life expectancy

Same battery 50% DOD makes 500 cycles - one cycle is 50Ah, totaling 25kAh.

Same battery 70% DOD makes 900 cycles - one cycle is 30Ah, totaling 27kAh.

In fact the difference between 300 cycles to 900 cycles is not 3 times longer life expectancy (300%), but only 3kAh or 12.5% more energy throughput.

in real world we have an average charge / discharge pattern and cycles cannot be measured / verified, but it is easy to count Ah drawn since installation day.

[CatNewBee]

It would be best to say, this AGM battery costs 150€ and delivers 25kAh or 300kWh (at 12V nominal) by moderate usage to 50% DOD.,

Considering it a consumable, the price for this AGM per kWh is 0.5ct (150€ / 300kWh)

Lets say a 100Ah LFP costs 800€ with BMS and delivers conservative 3000 cycles at DOD 20% until it declines to 60% of nominal capacity.
Total Energy would be 80Ah x 3000 cycles = 240kAh or 3120 kWh (at 13V!)
The price per kWh for this LFP would be 800€ / 3120kWh = 0,25ct

So even the LFP is 5 times more expensive upfront, it's usage is half the price of AGM power.

[nebster]

Originally Posted by john61ct

Problem is, both the 100% starting point, and the AH (mAh) ratings are completely arbitrary for each vendor.

Wait, what? The paper is talking about measuring the starting and ending points using the chemical potential directly, which is exactly the standard we should all be using. (It is certainly the standard that most others use.)

Absolute capacity for this paper's purpose -- and, indeed, for potential-based discharge stopping strategy that we can apply to our own systems -- is irrelevant.

So the specific results will vary not just by model / brand but by production date.

No, they won't, if it's the same LiFePO4 chemistry.

Anyway, from a practical POV there is no reason to **ever** allow your bank to go below say 11.9V in daily cycling.

Agreed, if you have a 4s LFP bank.

This paper makes clear what many have observed in the past: the chemistry is actually pretty resilient and can tolerate an additional degree of over-discharge, at least once or twice.

This is why bottom balancing makes me nervous.

Again, huh?

The whole point of a bottom balanced series string is to reduce the real issue at deep discharge: that the level of deep discharge won't be even! To use the paper's terminology, if you took a 4s string down to "1.05DOD", you may well have one cell in that string at the equivalent of "1.10DOD" and another one at "1.00DOD".

So in that scenario, you might have one cell that is fatally damaged while the other cells can recover from the (one-time) mistake.

Top-balanced/active-balanced packs all suffer from this problem at deep discharge, and for those of us who are more confident that we can stop charging than that we can stop discharging, this issue is basically the entire justification for bottom balance.

[CatNewBee]

Originally Posted by nebster

Wait, what? The paper is talking about measuring the starting and ending points using the chemical potential directly, which is exactly the standard we should all be using. (It is certainly the standard that most others use.)

Absolute capacity for this paper's purpose -- and, indeed, for potential-based discharge stopping strategy that we can apply to our own systems -- is irrelevant.



No, they won't, if it's the same LiFePO4 chemistry.



Agreed, if you have a 4s LFP bank.

This paper makes clear what many have observed in the past: the chemistry is actually pretty resilient and can tolerate an additional degree of over-discharge, at least once or twice.



Again, huh?

The whole point of a bottom balanced series string is to reduce the real issue at deep discharge: that the level of deep discharge won't be even! To use the paper's terminology, if you took a 4s string down to "1.05DOD", you may well have one cell in that string at the equivalent of "1.10DOD" and another one at "1.00DOD".

So in that scenario, you might have one cell that is fatally damaged while the other cells can recover from the (one-time) mistake.

Top-balanced/active-balanced packs all suffer from this problem at deep discharge, and for those of us who are more confident that we can stop charging than that we can stop discharging, this issue is basically the entire justification for bottom balance.

A BMS, what does bottom balancing also takes care of under voltage protection on the cells. It will disconnect the loads first and then discharges all cells to the pre-set bottom voltage, then it will eventually start charging of the bank.

And this is the disadvantage of bottom balancing. You have to wait until all cells are discharged and re-charged at least to some point, before you can use the battery again, this can take a long time on large banks.

Active balancing at the bottom can mitigate this disadvantage by transfering current from the highest cell to the block before one cell hits the bottom.

[nebster]

Originally Posted by Maine Sail

This study was charge to 4.0VPC or 3.8VPC and stop charging immediately upon hitting the target voltage and the discharge load turns on.

My questions are always:

* Why push into the vertical when it's not needed for capacity?

* What benefit is there by pushing into that area where there is really no quantifiable extra capacity?

* What would the outcome on cycle life be if you did not push into that knee every cycle, even with a "stop charge" protocol?

Yes, this, exactly.

I think the answer to your first question is "because the people doing the research are inexperienced in the field," which is fine. Everyone has to start somewhere.

But what we really would benefit from is understanding more about longevity versus more conservative voltage stopping points. (And, while we're dreaming, versus a range of temperatures, and versus more varied PSOC than simply linear up-down-up-down bouncing between the two endpoints.)

Practically speaking, though, I think we all can agree that it feels like staying between 3.0V (~4% SOC at 0.2C), and 3.45V or 3.50V (~89-92% SOC at 0.2C) has got to be fairly close to optimal for those of us okay with 80% left at seven to ten years out.

[CatNewBee]

@ Nebster: The theory of bottom balancing is something else.

The main assumptions are

- one of the cells is weaker than the others, so it hits the bottom first on discharge and will hit the roof first on charge (less capacity)
- cells live longer when stay at lowest possible SOC.

So when the battery is discharged and first cell hits the bottom, you stop discharging of the battery and discharge all other cells to the same voltage as the lowest one (bottom balancing). then you charge to full (first cell - assumably the same - reaches the cut-off top voltage. All other cells have the same energy charged as the weakest at lowest possible cell voltage. The claim is, they will live longer because kept well below maximum charge.

In the real world cells have slightly different inner resistances, that lead to different charge / discharge efficiency and so different temperatures of the cells, and different current absorption. Some cells store more energy quicker than others in the string, there are voltage differences on the cells while charging, so the assumption is not always valid.

It is just another strategy to get a somehow balanced battery and maximize the usable capacity.

Main disadvantage is the downtime during balancing and re-charging to a usable SOC before you can continue to use the battery.

[nebster]

Originally Posted by CatNewBee

A BMS, what does bottom balancing also takes care of under voltage protection on the cells. It will disconnect the loads first and then discharges all cells to the pre-set bottom voltage, then it will eventually start charging of the bank.

I've never heard of a BMS that "does bottom balancing." Where have you seen one of these?

What I assumed john is referring to, and what most everyone means when they write "bottom balancing," is the process of MANUALLY balancing all the cells, ONE TIME, at the start of provisioning a new pack. And then leaving them alone and charging/discharging them all together.

There is no active, cell-level charge management in a "bottom-balanced" pack.

And this is the disadvantage of bottom balancing. You have to wait until all cells are discharged and re-charged at least to some point, before you can use the battery again, this can take a long time on large banks.

That sounds crazy. Maybe it is why no one does it.

[nebster]

Originally Posted by CatNewBee

@ Nebster: The theory of bottom balancing is something else.

I don't know what this sentence means. What I do know is that, when you on a forum where people are discussing "bottom balancing" a battery pack, they mean "we will voltage-align all the cells at the bottom of their charge."

- cells live longer when stay at lowest possible SOC.

This is generally not a motivation for bottom-balancing as the term is commonly used. The reason we bottom balance is because it provides an extra degree of safety in the event of a low voltage condition.

So when the battery is discharged and first cell hits the bottom, you stop discharging of the battery and discharge all other cells to the same voltage as the lowest one (bottom balancing). then you charge to full (first cell - assumably the same - reaches the cut-off top voltage. All other cells have the same energy charged as the weakest at lowest possible cell voltage. The claim is, they will live longer because kept well below maximum charge.

I understand what you are writing, and I am writing back that I'm not aware of anyone actually doing what you are talking about.

In the real world cells have slightly different inner resistances, that lead to different charge / discharge efficiency and so different temperatures of the cells, and different current absorption. Some cells store more energy quicker than others in the string, there are voltage differences on the cells while charging, so the assumption is not always valid.

While there are no doubt bad cells that don't perform well in some cases, I can tell you that there are a lot of lithium iron phosphate packs that are running just fine with ONE SINGLE bottom balance at the start. The tiny theoretical variations you describe are simply not a factor in the real world packs that some of us are using.

Main disadvantage is the downtime during balancing and re-charging to a usable SOC before you can continue to use the battery.

Again, you are writing about something that

(a) no one does in the real world, today, that I've ever seen

and

(b) is not the commonly accepted meaning of the phrase "bottom-balanced battery."

I am correcting you here again so that it is clear(er) for other readers. You are free to define the terms in your head to mean whatever you like.

[CatNewBee]

Initialization of cells is usually done by the initialisation charge to 4.0V before first using them - this is what the manufacturer says and what you need to do to unleash the full potential of the LFP. This is a TOP balancing approach.

There are "DIY"- BMS and circuit plans in the wild advocating bottom balancing, but there are no commercial builds for the obvious reason relying sole on bottom balancing. Most of the DIY circuits are semi-automatic (adding cell loads, that shut at a given voltage) or manual (opening the cell bridges, connecting the cells in parallel and discharging the pack to a given voltage, then rebuilding the battery for charging).

The idea, however is not dead in commercial products. There are manufacturer, who do active balancing at all SOC stages - like the ABMS from REC. Their philosophy is to keep the cells always perfectly in balance and get most juice out of them even if one cell starts dying - until it gets replaced - and this with less possible power loss.

What they do - is:

- actively transfer capacity on the top when charging from the highest cell to the pack (TOP BALANCING), there are threshold voltages and amps when balancing starts and when it stops - completely configurable,

- monitoring cell voltage during discharge and invoking cell balancing + rising a warning if the voltage differs more than a preset acceptable configurable threshold by charging the lowest cell by the pack (ERROR MITIGATION).

- actively rescuing the lowest cell in case low voltage protection kicks in and the voltage difference to the other cells exceeds a configurable threshold, that allows power transfer from the pack to the lowest cell (BOTTOM BALANCING) , then re-enabling further discharge, so the user gets more capacity until he can re-charge the whole shebang.

They do the balancing by DC-DC power converter, instead of burning capacity by resistors.

All 3 strategies can be tweaked or suppressed by configuration setting.

[Delfin]

Originally Posted by ColdEh Marine

I have just such a system .

I charge to 14.2 volts usually with no current tail off . When I hit this voltage I simply turn off my generator .

When my system hits 12.8 volts with no load I start charging again . I have found this method keeps me within the knees and I have yet to experience any cell drift. Three years in of full time use. My system generates on average 200 amps when charging and the generator now has 200hrs on it. I have 600 amp hrs of Sinopoly cells on board .

Regards John.
www.coldeh.com

Very, very similar to our system and usage, although we generally re-charge before hitting 12.8 volts, just because it fits a daily pattern, and I disconnect charging at 14 volts (28 for me).

One question I have, and this goes to Tanglewood's question about research showing that a shallower depth of charge increases cycle life, stems from the data Lithionics has posted here.

If you look at how they present the data, they have tested to 2,700 cycles at 80% discharge, then, it looks like they are just extending the rate of cycle increase to predict 35,000 cycles at 10% DoD - what they identify as "analog result".

However, if you look at this carefully, they are defining x% DoD in what looks like an odd way to me. 50% DoD is defined as ranging between 75% and 25% SoC, not 95% to 45%. This would suggest that they think that in addition to increasing cycle life by reducing DoD per cycle, you also would benefit from operating that cycle within the "middle" of the power curve, if you understand what I mean. Am I processing this correctly?

[Delfin]

Originally Posted by tanglewood

Just curious.. what research is there that shows thousands of more cycles when using lower depth of discharge?

The Papezova paper would indicate that (table 1 vs. table 2). One wonders whether the cycle life would not have been greater if the charge voltage was lower than the 15+ volts they used in the lab....