Best way to manage this LFP house battery

[karrak]

Okay, here's another attempt at the same attachment. Let's see if it works better.

Your graph is showing up now.

Now my graph is showing up as the broken picture icon if I am not logged in and I clear my web browser cache. Shows up OK if I am logged in???

Edit: Deleted graph and reloaded and it is no longer showing the broken picture icon. Just back to showing the name of the picture. This seems to be the way the Solar Panel forum website is set up. Very annoying that you have to be logged in to see the pictures.

Simon

[PNjunction]

I'm assembling a new LFP for my RV out of Chinese cells. These are 100Ah prismatics, and I have 112 of them for 16s7p. My plan is to add isolators for each 16s string so they can be pulled out for service or disconnected automatically if something is amiss.

STOP RIGHT NOW.

That's not how you do it, or you will *never* maintain balance, either passive or actively.

With LFP, here's how you do it:

1) Parallel cells to build up your desired capacity first. Cells should be matched close to capacity and internal resistance.
2) Series connect these paralleled-cell groups to achieve your desired voltage

The next problem I see is that you are bottom balancing, yet you are talking about absorb and end-currents. With bottom balancing, you stop charge when the first cell reaches your pre-defined target voltage. Are you set up to monitor 112 cells to act as this trigger?

What you are describing now is a system that is trying to do BOTH top and bottom balancing and you'll get the worst of both worlds.

You have to choose only one method, top or bottom. Unless you have the skills and gear to manage it properly so that you don't inadvertently end up doing both top and bottom, I recommend buying 4 more cells and get in some badly needed practice:

https://www.solarpaneltalk.com/forum...prismatic-bank

Please get some hands-on with a learner-battery first.

Since those cells may be random shelf-picks, you are looking at spending some loving quality time measuring them for both capacity and internal resistance, to make sure your paralleled groups are closest to each other. A Revolectrix PL-8, or other high-end quality charger/analyzer is called for.

[max2k]

Your graph is showing up now.

Now my graph is showing up as the broken picture icon if I am not logged in and I clear my web browser cache. Shows up OK if I am logged in???

Edit: Deleted graph and reloaded and it is no longer showing the broken picture icon. Just back to showing the name of the picture. This seems to be the way the Solar Panel forum website is set up. Very annoying that you have to be logged in to see the pictures.

Simon

your pic was never showing up for me, I just didn't mention it.

[nebster]

I think the point is when monitoring string voltage of 16 cells the whole 'swing' is 0.6V x 16 = 9.6V which can bury multiple cell voltages in it- you can have 10 x 3.5 + 6 x 3.2 vs 16 x 3.39V and many other combinations and they will be indistinguishable from the total string voltage point of view. You probably can mitigate this by measuring parts of 16 cell string, like every 4 cells. Supposedly LFPs don't drift too far from each other but again this has been reported by only handful of people. jflorey2 implies the opposite- his bank got weaker cell (#13 naturally) he needs to watch over closely.

Yeah, I think that's right -- one individual cell will always spike higher than the others. It reaches 100% SOC faster and puts the string at risk if charging continues.

The antidote is to identify the string voltage at that high point and commit to never raising the string above that (combined) voltage.

I believe the questions, then, are:

1) Is the least-capacity cell, when full, still big enough to allow for an acceptable total charge? (Certainly yes, in my test pack's case.)
2) Will the least-capacity cell retain its same capacity relative to the rest of the cells in the string over time? If yes, then the pack-level voltage stopping point will work fine. If no, then we need to monitor and/or actively rebalance and/or replace the troublemaker cell.

Anecdotal evidence gleaned from others with large long-term LFP installs syggests that the relative capacities between cells remain very stable over time. My theory is that, if these cells did drift a lot, we would hear a lot more about outright pack failures in conversation about it on various forums. But... who knows?

[PNjunction]

nabster ..

Anecdotal evidence gleaned from others with large long-term LFP installs syggests that the relative capacities between cells remain very stable over time. My theory is that, if these cells did drift a lot, we would hear a lot more about outright pack failures in conversation about it on various forums. But... who knows?

I think we're all typing at the same time.

Heed my warnings above about your idea of string wiring. That is lead-acid think. Not applicable here, and you'll never *maintain* balance.

Most of these mistakes were already made and discussed on the EV forums about 10 years ago. That is why you don't see much of them today.

[nebster]

Thanks for the emphatic suggestions, Mr. PNjunction! I fear we will have to agree to disagree on at least few things.

The actual LFP packs in actual boats and RVs, right now, are in fact generating anecdota that overwhelmingly suggests:
1) active balancing/shunting is error-prone and causes more problems than it solves
2) capacity drift is quite low, cell to cell, even after years, in a stable, well-managed pack
3) layers of conservatism yield predictable results (not unlike many of life's endeavors)

Parallel-then-serial arrangements are nice when system downtime is more tolerable and when maintenance is always readily available to mitigate a problem. They are still sensitive to capacity-balance and -drift, just like s-then-p topologies. There are tradeoffs, and there is most definitely not one way right way to wire a pack. Almost all engineered LFP solutions end up as a serial-than-parallel arrangement and, presumably, work well. I have chosen s-then-p, too, because I want the ability to isolate an ailing cell and keep the rest of the system energized. I am aware that this subjects the entire 112-cell (unbalanced) pack to the capacity constraint of the worst cell in the whole bunch. If that ends up being a problem, I will have to reassess and possibly even switch.

With my bottom-balance strategy, my plan was to bring the string up until the "worst" cell hits 3.6V. However, the first few cycles have hit the CV threshold before the worst cell reaches 3.6V, so I have been in CV for a period of time afterwards. In the run graphed above, it took about 30 minutes of additional CV to hit the 3.6V. It takes me a while to do a manual load cycle, so only every few days do I have a chance to run a charge at a different rate or with a different CV threshold. It's fun to experiment with, but it's also a bit tedious!

My original open question is still: can I achieve a reasonable charge routine for this pack with only a CC current limit, a CV threshold, and the ability to stop CV after 1+ hours? I think the answer is... probably.

[nebster]

Heed my warnings above about your idea of string wiring. That is lead-acid think. Not applicable here, and you'll never *maintain* balance.

I'll have to ask you to defend your last clause a bit. I guess we could start with you being very precise about what you mean by "balance"?

What I am interested in is whether relative cell capacity drifts over time. The data I have access to -- and it is not much, but I have to go with I do have until I see otherwise -- suggests that people are observing high relative capacity stability over time. If you can show me otherwise, I'm all ears.

[max2k]

nabster ..



I think we're all typing at the same time.

Heed my warnings above about your idea of string wiring. That is lead-acid think. Not applicable here, and you'll never *maintain* balance.

Most of these mistakes were already made and discussed on the EV forums about 10 years ago. That is why you don't see much of them today.

IMO it's lead acid approach to connect 7x 100Ah LFP cells in parallel- anything goes wrong in there the fire is guaranteed. It is pretty much guaranteed even with 1x 100Ah cell I think but might be x7 times easier to contain. I came across interesting post on the topic: https://electronics.stackexchange.co...-battery-packs while not related to LFP directly it outlines few approaches Tesla took to increase safety of their batteries. They actually connect a lot Li-ion in parallel but if my understanding is correct their idea of safety is to use small cells and contain single cell failures. These 100Ah LFP cells have much lower resistance and bigger size making it harder to 'contain' as in case of shorting remaining parallel cells will readily supply any current.

BTW, LFP in their current shape only started to appear or didn't even exist 10 years ago so we don't really have enough empirical data. Those older banks only today begin to present interesting study subject. I'm just trying to be on a safe side and assume short will happen, fire will start and design for that.

[PNjunction]

If you read my KISS setup, you'll actually see I am a proponent of simpler methods, like not having a .01C balancer mosfet fail closed, and now unbalancing a $5K bank!

Leaving out the EV threads, I'm assuming you took a month's vacation to go through this thread - although at this stage, it may be better to read it backwards!

http://www.cruisersforum.com/forums/f14/lifepo4-batteries-discussion-thread-for-those-using-them-as-house-banks-65069.html

And of course this one:

http://www.pbase.com/mainecruising/lifepo4_on_boats&page=1

I'm just saying many have spent good money going down the path, but strayed too far with empirical evidence and logic, and are nowhere to be found today. There are plenty of "one-off's" that initially seem to be operating great with unconventional setups, but where are they today? Did they achieve the >2K promised cycles treating LFP like lead-acid?

I'm just saying - follow the pioneers that are still around with over 10 years experience, not the prognosticators and marketers trying to lead you to believe that LFP is a "drop in" for lead-acid in either construction or technique. Oh you'll get away with it early on. But man, down the road is another reality.

[PNjunction]

I'll have to ask you to defend your last clause a bit. I guess we could start with you being very precise about what you mean by "balance"?

In your proposed arrangement of swappable strings, during use one string will do more work than the other. This leads to an overall imbalance and eventual failure.

With all the cells at your disposal, you can prove it to yourself by building a learner-bank the wrong way. Watch what happens. I have.

This is a major problem with LFP discussion on many boards. Far too many with generally helpful participation, but far too few with actual hands-on experience. (and no, benchtop hacks of trashy laptop cells don't count.)

I'm just trying to save you some time, effort, and money. Although quite a few of us just don't have the energy for the endless debates any more.

[max2k]

In your proposed arrangement of swappable strings, during use one string will do more work than the other. This leads to an overall imbalance and eventual failure.

How 2 parallel strings are different from 2 parallel cells in this regard? 'Weaker' (less C) string would be charged sooner but the 'stronger' string will be holding voltage back still taking up the charge so the currents will be split correctly between them. The same would happen during discharge. This is of course if they were brought to the same voltage (within 10mV) before connecting in parallel otherwise very high 'balancing' current will flow into less charged string possibly damaging it.

With all the cells at your disposal, you can prove it to yourself by building a learner-bank the wrong way. Watch what happens. I have.

It would be great if you could share details of your experiments, that might spare bunch of us a lot of time/money.

[sensij]

It would be great if you could share details of your experiments, that might spare bunch of us a lot of time/money.

Here is one thread, for example.

https://www.solarpaneltalk.com/forum/off-grid-solar/batteries-energy-storage/17106-parallel-battery-connections

[karrak]

Thanks, I haven't considered going with a lower CC rate, but that's mostly because I anticipate using the genset to charge underway, and I'd like to run that at as high a load for as short a time as possible. That is in conflict with my desire to extend absorb times past 1 hour, though. I don't yet have a mechanism to stop CV beyond a crude timer with a minimum of 1 hour. I've definitely got to work that out!

Can you tell us the make and model of all the equipment you will be charging your battery with and the different charge rates you will be using, is solar anywhere in the mix?

What do you think about only monitoring a handful of the "troublemaker" cells? So far it seems completely consistent: the same couple cells are the ones with presumably the least capacity, so they spike up much earlier than the rest. I'd really like to avoid rigging up 100+ monitoring circuits if I can help it.

The problem I see with this is that the problem cell in five years time might be a good cell at the moment. I know of one case from a knowledgeable and credible poster on the Australian Energy Matters forum of a CALB cell loosing around 25% of capacity after about three years of use. It was his BMS that picked this up. I have not heard of any cases of cells developing internal short circuits without being abused beforehand. As far as I am aware (beware I am not an electrochemical engineer) the two ways that cells can develop internal short circuits are

1. Cell being badly over discharged which results in the copper current collector which is part of the battery cathode being oxidised. If a cell in this state is then charged up again the copper can be reduced back to copper but instead of the smooth copper sheet that it was it can grow pointy copper crystals that can grow between the cathode and anode and eventually short the cell out. These crystals (dendrites) do not have a zero electrical resistance. If there is enough energy in the cell with the dendrite and other cells in parallel and the dendrite resistance is low enough the dendrite will be vapourised and hopefully remove the short. A more dangerous scenario is if there is not enough energy to vapourise the dendrite but enough to heat it up to the point where the other components in the cell around the dendrite start to decompose and add to the heat output. Another issue that could cause problems is that the resistance of the sections of the copper current collector that have been eaten away by the oxidation could become high enough to cause these sections to heat up enough to cause other parts of the cell to decompose.
2. Cell being held at a high SOC and high voltage for a long period of time which can lead to lithium dendrites growing from the anode. This takes some time and I don't think is as likely to cause a catastrophic failure of the cell.

I think it is far more likely that over the ten plus years lifespan of the battery that the chance you will get a cell overcharged or over discharged because of changes in individual cell capacities relative to each rather than problems with any cell spontaneously short circuiting.

With the two systems I am responsible for which are organised with cells in parallel and then is series I have found that I only have to turn the system power off for a few minutes if I need to do any work on the battery. Even if you have the complete failure of one cell in the battery it would still work but with a reduced capacity.

I was planning to observe the system and deduce when a rebalance was necessary. It would be unacceptable to have to do it every few months, but once or twice a year would be okay. I expect we will cycle the pack roughly twice a week in real life.

I know there are (unmonitored) bottom-balanced LFP packs staying very well balanced after several years in the field, so I have been assuming (hoping) that mine would behave similarly.

The main reason I don't like bottom balancing for off-grid systems that are in constant use is that the only way you know if the battery has stayed in balance is to empty it. This might be fine for a EV sitting in a garage for the majority of its life but is not very practical for systems in use all the time. If you top balance you know every time you charge the battery if it has stayed in balance. By charging the battery to nearly 100% you can also easily and accurately reset your SOC meter. I have found my coulomb(current) counting SOC meter is the most useful way of measuring the state of my battery and allows me to plan my energy usage.

Simon

Off grid 24V system, 6x190W Solar Panels, 32x90ah Winston LiFeYPO4 batteries installed April 2013
BMS - Homemade Battery logger github.com/simat/BatteryMonitor/wiki
Latronics 4kW Inverter, homemade MPPT controller

[nebster]

In your proposed arrangement of swappable strings, during use one string will do more work than the other. This leads to an overall imbalance and eventual failure.

I agree with every word you wrote, and by the way, it also must be the case that this is true at the cell level with parallel-then-serial. In the latter the failure modes do manifest somewhat differently.

The question, though, is how much imbalance and when failure? If it's gone pear-shaped after ten charge cycles, that's a total suck and a non-starter. If it takes 300 cycles, I'll gladly do some checkpointing and swap cells if needed between strings a few times a year. (It's easy to do with my topology!) If it takes 1000 cycles, I'll be nearly dead by then and don't care too much.

With all the cells at your disposal, you can prove it to yourself by building a learner-bank the wrong way. Watch what happens. I have.

...

I'm just trying to save you some time, effort, and money. Although quite a few of us just don't have the energy for the endless debates any more.

I get that debate can be tiring, and I truly appreciate your good intention to warn me... but, really, I'm mostly interested in data. Can you point us to some? It sounds like there is some maybe even stored away in your own brain/build logs!

The limited set of applicable data I have been able to source suggests that the imbalance will be manageable. My own crude mathematical model also suggests that it should be doable. I would be happy to be proven wrong before I torque another 224 nyloc nuts.

[karrak]

I mean, I see what you're saying, but I think there's ample room to disagree: far easier for me to run my meter down the cells on an appropriate interval than to wire in hundreds of leads and terminals and add Arduinos or sketchy PCBs with unbalanced draw (I'm looking at you, celllog8). If the cells are stable with respect to each other in a nominal sense, then real-time complete monitoring seems like overkill. I really like simple with critical systems.

Cell imbalance only shows up when the battery is nearly fully charged or discharged, so you will have to wait for the right time and them take 112 readings. The cellog8 has balanced draw if you modify it so that it is powered by an external power supply.

BalVolts&ChargeCurrent.jpg

Simon

Off grid 24V system, 6x190W Solar Panels, 32x90ah Winston LiFeYPO4 batteries installed April 2013
BMS - Homemade Battery logger github.com/simat/BatteryMonitor/wiki
Latronics 4kW Inverter, homemade MPPT controller