Run parallel cells and string them together, or run independent strings and parallel them?

[ryandc]

So, a little bit of background on me - I've done a lot of construction work over the years (put myself through college doing apartment rehab and appliance repair), and have done a number of small-fry solar installations (usually just remote 12v field site stuff done in conjunction with mesh wireless). So I'm familiar with solar and electrical principals and concepts. But - my day job was scale-out storage. So I'm used to thinking of things in terms of mean time between failure, failure points, and fault mitigation.

Which lead me down the path of a lot of googling, that unfortunately has lead me to mixed reviews. This is my first post here, and this forum seems to be the nexus for most of the subject matter expertise regarding LifePo4 battery banks out there.

I'm aiming to build a 48kwh 48v LifePo4 bank. As I see it, there are 2 high-level ways in which I can do this.

Method 1: Parallel the cells
In this method, I'd take 3 320ah 3.2v LifePo4 cells or 4 250ah 3.2v LifePo4 cells and parallel them, then connect each paralleled unit to the next, and so on.

Cons:
- Would require higher amperage bus bars
- In the event of any failure of any individual cell, the lifetime of the other two to three cells in parallel would be shortened due to depth of discharge increasing for the other cells in parallel

Pros:
- Single BMS
- Failure of any single cell would result in higher draw on cells in parallel, but would not result in substantial loss of capacity for the bank (discharge, yes, capacity, no)

Method 2: Parallel the strings
In this method, take 16 320ah 3.2v LifePo4 cells and string them together. Do that with three strings. Parallel the strings

Cons:
- Any fault in any individual cell will effectively take the string offline due to voltage drop
- Would require multiple BMS


Pros:
- Can readily identify faulted cell (in paralleled cells, you can identify quickly which set of cells are faulted, but not the individual cell)
- Because of how the faulting is handled (namely whole string goes offline), no long-term damage to the string. Swap out cell, back online
- Higher voltage / less amperage = fewer bus bars

You can mix and match as well, of course. Build a few 12v strings, parallel those, then link them to create a 48v string. Etc.

I've searched a long time to determine what the best practice here is - good chance this has been discussed elsewhere and I'm using the wrong search terms. Anyone have any recommendations as to best practices here?

 

[sunshine_eggo]

Method 1: Parallel the cells
In this method, I'd take 3 320ah 3.2v LifePo4 cells or 4 250ah 3.2v LifePo4 cells and parallel them, then connect each paralleled unit to the next, and so on.

Cons:
- Would require higher amperage bus bars

Not usually. BMS is still the limiting factor. You're actually sharing the same current between multiple cells with more bus bars. More bus bars are required, but they don't need to be higher amperage.

- In the event of any failure of any individual cell, the lifetime of the other two to three cells in parallel would be shortened due to depth of discharge increasing for the other cells in parallel

Correct, but see Pros

Pros:
- Single BMS

Generally a pretty small portion of the total battery cost, so not a huge pro.

- Failure of any single cell would result in higher draw on cells in parallel, but would not result in substantial loss of capacity for the bank (discharge, yes, capacity, no)

Optimistic. The failure of a parallel cell often renders the battery mostly unusable due to the severe cell variations induced. Premature discharge limits and premature charge limits.

Method 2: Parallel the strings
In this method, take 16 320ah 3.2v LifePo4 cells and string them together. Do that with three strings. Parallel the strings

Cons:
- Any fault in any individual cell will effectively take the string offline due to voltage drop

True, but this is also a Pro as you identified.

- Would require multiple BMS

Yep, which often allows the total bank current to be higher.

Pros:
- Can readily identify faulted cell (in paralleled cells, you can identify quickly which set of cells are faulted, but not the individual cell)

Correct

- Because of how the faulting is handled (namely whole string goes offline), no long-term damage to the string. Swap out cell, back online

This also supplies limited redundancy. Simply disconnect the bad battery from the bank, and you're still operational.

- Higher voltage / less amperage = fewer bus bars

Still BMS limited, so not less amperage, but the two separate batteries do require fewer bus bars.

2P16S: 47 bus bars minimum
16S2P: 32 bus bars minimum

You can mix and match as well, of course. Build a few 12v strings, parallel those, then link them to create a 48v string. Etc.

If you're going to make 48V, you should do 16S, not 4S * 4. that gets expensive as you need a BMS for each 12V, and each 12V BMS needs to be able to handle > 48V on the electronics level, or they'll be damaged .

I've searched a long time to determine what the best practice here is - good chance this has been discussed elsewhere and I'm using the wrong search terms. Anyone have any recommendations as to best practices here?

More reading:

Cell Configurations for 12V 24V and 48V LiFePo4 Batteries

This short deck shows 1P and 2P cell configurations for 12V, 24V & 48V LiFePO4 batteries. To get the deck, click on the orange button at the top of this page. Note: Reviews, Comments and corrections are welcome Document Revision History...

diysolarforum.com

 

[acdoctor]

I advocate for 1 cell per bms sense wire. My opinion is 1 bad cell in parallel with 2 others that are healthy, and 1 sense wire the voltage will be kind of averaged and the bms thinks it’s okay. When in reality 1 cell needs your attention now!

 

[BentleyJ]

This issue also comes up when stacking inverters. For example, say you have 3 inverters connected in parallel to increase output. Would you build 1 monster battery pack or 3 smaller packs, one for each inverter but connected on a common bus bar. My answer: match the system based on points of failure. If you only have one Inverter then one battery is probably OK, if you have 2 inverters stacked then I would go with 2 batteries. Etc. Makes sense to me anyway, what has the higher probability of failing a battery cell, the BMS or the Inverter. I don't know, I've not seen any statistical data on the issue. Perhaps only because I haven't really searched for it.

 

[sunshine_eggo]

This issue also comes up when stacking inverters. For example, say you have 3 inverters connected in parallel to increase output. Would you build 1 monster battery pack or 3 smaller packs, one for each inverter but connected on a common bus bar. My answer: match the system based on points of failure. If you only have one Inverter then one battery is probably OK, if you have 2 inverters stacked then I would go with 2 batteries. Etc. Makes sense to me anyway, what has the higher probability of failing a battery cell, the BMS or the Inverter. I don't know, I've not seen any statistical data on the issue. Perhaps only because I haven't really searched for it.

I've never seen a manufacturer allow for multiple parallel inverters to be run from separate battery banks. They always indicate they should be powered from the same battery bank.

Most installation instructions point out the importance of having equal length cables for BOTH inverters on both DC and AC side. Imagine how useless that would be if you were using completely separate batteries for the inverters - they'd be pulling different current based on the different battery voltages.

 

[time2roll]

I built mine 2p4s2p to have a balance of the best and worst parts of the equation.

 

[toms]

Parallel first then series, much simpler = more reliable.

Batteries in parallel inevitably have balancing issues, if you go that way make sure you leave access to individually charge each battery in parallel.

If you are concerned about redundancy, use multiple seperate entire systems.

It’s a misconception (that i’ve only ever seen on this forum) that parallel cells can somehow have different voltages. The concern is an internal (dendrite growth) short will damage all cells in parallel - you need to ask yourself where this information is coming from, i’ve not seen or heard of it happening. If there is an external short (ie dropped spanner) or BMS failure then you may lose all cells in a parallel group.

I’ve commissioned hundreds of cells in parallel over many years without issue.

There is plenty of information on this forum regarding parallel or series first - not much long term experience.

 

[BentleyJ]

I've never seen a manufacturer allow for multiple parallel inverters to be run from separate battery banks. They always indicate they should be powered from the same battery bank.

Most installation instructions point out the importance of having equal length cables for BOTH inverters on both DC and AC side. Imagine how useless that would be if you were using completely separate batteries for the inverters - they'd be pulling different current based on the different battery voltages.

If you would have read my post completely, you would have noticed, I SPECIFICALLY stated the 3 batteries would be connected together via common bus bar thus the inverters would "see" only one battery.

 

[sunshine_eggo]

I read it. I just didn't want someone else reading it to think it was an option.

 

[Sanwizard]

My opinion is seperate 16s batteries, each with its own BMS in parallel. You can turn off an entire battery for maintenance, while staying online. You can get the health of every cell individually, so you will catch bad cells faster, and with less impact.
It costs a bit more for the extra BMS's, but you have great redundancy.
This reminds me of the mirrored stripes or striped mirrors debate in RAID storage.