Parallel cells - what are the pitfalls as the number of cells in parallel gets deeper ?

[Ohms_Cousin]

Have read read many posts here re paralleling cells. Not seen anything really in depth about it though.

I get that a group of cells that are paralleled all have the same voltage and that you lose visibility on each individual cell.

What I am trying to figure out is what is the practical limit to stop paralleling cells from a safety standpoint? A 2P paralleled cell set seems to be fairly low key in terms of safety. With only 2 cells and an avg voltage there doesn't seem to be too much that could go astray.

But how does it look at 4P or 5P or 6P ? How much more risk does one take on for every extra paralleled set of cells?

My question relates to getting enough capacity in kWh without using too many paralleled batteries in a bank. I came across a cylindrical cell that is 3.2v 22Ah. That is only 70.4 Wh per cell. At 16s that is only 1.126 kWh in that battery, so would take a lot of paralleled batteries to get to my target of around 50kWh.

Seems it would make more sense to go 4P16s = 4.5 kWh in one battery - so still needing 10 or 11 batteries to get up to 50kWh.

If one were to go 6P with the 3.2v 22Ah cells (6P16s) then it is 6.7kWh per battery and cuts number of batteries down to about 7 or 8.

The more cells one puts in parallel the less batteries needed. Depending on BMS chosen this makes a big difference in BMS costs.

Other factors are getting even charging into so many batteries. I read here that to get even charging one should stick to either 2 or 4 batteries in parallel. Seems to be a lot to consider in all this.

Why cylindrical cells? I want to build a battery for a boat. I have read at several sources that bigger prismatic cells can be trouble in a boat and that one should stick to around 100Ah if using prismatic cells. Seems the best way to avoid the whole is to use cylindrical cells - they offer better cooling with more surface area exposed to air and should overall be more robust because of shape.

I have read how in parallel the cells are all the same voltage and that one loses visibility on individual cells and understand that. However how big an issue is this at 4P or 6P? Is there a point that it becomes too big a risk and its better to go no more then, say, 4P and just put more BMS's into a system? Or is this not a thing at all to be concerned about?

 

[BradCagle]

I dunno, but for what it's worth I have a 16s pack that has 25x 5ah cells in parallel

I guess the risk is higher odds that one of those cells will be defective, and take out the whole section.

 

[sunshine_eggo]

Tesla is cool with something like 86 NCA cells (the ones that can burn violently/explode) in parallel, so YMMV.

It's probably more about redundancy. If a single cell goes out rendering that whole battery inop, how little of the original battery can you live with before usability is a concern as you troubleshoot the failed battery? If you can live with 75% of it, shoot for a XP16S configuration that results in 4 batteries in parallel.

Also, buy spare cells now.

 

[Ampster]

What I am trying to figure out is what is the practical limit to stop paralleling cells from a safety standpoint?

I am not aware of any safety limits with LFP as far as paralleling cells. Tesla uses a more volatile chemistry and has over fifty cells in parallel. Tesla does fuse each cell and has a cooling mechanism.

 

[Sverige]

But how does it look at 4P or 5P or 6P ? How much more risk does one take on for every extra paralleled set of cells?

<snip>

Other factors are getting even charging into so many batteries. I read here that to get even charging one should stick to either 2 or 4 batteries in parallel. Seems to be a lot to consider in all this.

What risk is it you’re concerned about? I can’t see any risk which increases as the number of cells in parallel is increased.

Also, you may have read it on here, but doesn’t mean it’s true - “getting even charging requires no more than 4 batteries in parallel” sounds like made up nonsense to me. Cells in parallel are at the same voltage by definition and will distribute charging current between them in proportion to their capacity (because if one cell took proportionally more, its voltage would tend to rise relative to the others, which cannot happen).

I think a lot of misinformation about paralleling cells is spread by people who have never done it, because they came up with some whacky theories when they thought about doing so and were then too afraid to try.

 

[Ohms_Cousin]

I dunno, but for what it's worth I have a 16s pack that has 25x 5ah cells in parallel

I guess the risk is higher odds that one of those cells will be defective, and take out the whole section.

You like to live on the edge huh.....

To me a 25P is a bridge too far. There have to be higher odds of a cell going bad on you when you times your chances by 25. At least it would seem so.

 

[Ohms_Cousin]

Tesla is cool with something like 86 NCA cells (the ones that can burn violently/explode) in parallel, so YMMV.

It's probably more about redundancy. If a single cell goes out rendering that whole battery inop, how little of the original battery can you live with before usability is a concern as you troubleshoot the failed battery? If you can live with 75% of it, shoot for a XP16S configuration that results in 4 batteries in parallel.

Also, buy spare cells now.

Yes but we dont know too much about how tesla manage these batteries and they have active cooling etc - and the odd fire here and there.

It depends on which BMS I decide to use. I like the Battrium approach with its Watchmon and K9's. The only real downside I see to that system is one contactor or shunt trip gets triggered and ALL batteries are brought down and offline. So that means 2 Battrium BMS systems so that I dont end up in the dark - at least a few batteries on a separate BMS to avoid total blackout if (when?) there is a trip.

The other way to do it which seems to be gaining traction here is a separate BMS for every battery -ie, a JK BMS for each battery - I presume that also means a contactor per battery. Still reading up on this trying to figure out how that all works - for example how does one see ALL the batteries at the same time? From what I can tell it would mean logging into each JK BMS to check cell voltages?

Then there is this BMS https://www.nuvationenergy.com/battery-management-systems/multi-stack-controller

It looks quite good but only came across this morning so lots of ground to cover on it yet. Still, it seems to cover all the bases. Its probably big $$ though. Still, 2 x Battrium BMS's is also $$.

Sticking to 4 batteries in parallel seems to be a thing - the Victron wiring guide shows 4 batteries in parallel. Must be a reason for that.

Yes, i plan to buy a good number of cells if I do end up going this route. I need to get the battery test gear and capacity match each and every cell and batch them together.

The Cylindrical cells are 5C discharge and 0.5C charge. So quite a robust cell.

 

[Ohms_Cousin]

I am not aware of any safety limits with LFP as far as paralleling cells. Tesla uses a more volatile chemistry and has over fifty cells in parallel. Tesla does fuse each cell and has a cooling mechanism.

is it a good idea to fuse LiFePo4 cells per each? I understand that its good practice to fuse each battery with a class T fast blowing fuse. That takes a battery off line if it goes wrong for sure, but what about at cell level? I know they do this with 18650 cells. Is it a good idea with LiFePO4 cells - they are 22Ah per cell Vs the 18650's being 3ish Ah per cell. Does that make any difference to fusing each cell at all?

 

[Ohms_Cousin]

What risk is it you’re concerned about? I can’t see any risk which increases as the number of cells in parallel is increased.

Also, you may have read it on here, but doesn’t mean it’s true - “getting even charging requires no more than 4 batteries in parallel” sounds like made up nonsense to me. Cells in parallel are at the same voltage by definition and will distribute charging current between them in proportion to their capacity (because if one cell took proportionally more, its voltage would tend to rise, which cannot happen).

I think a lot of misinformation about paralleling cells is spread by people who have never done it, because they came up with some whacky theories when they thought about doing so and were then too afraid to try.

The risk I am concerned about is the potential for a cell to go "bad" and be able to see it. It would seem that the more cells in parallel the more chances that something could go astray. Without being able to see each and every cell in a battery it seems to be a leap of faith that all the cells in a deep "P" battery is operating OK.

Re the 4 batteries in parallel and even charging. I saw a post here referencing the Victron wiring guide and Victron show 4 batteries being paralleled. If this is not some sort of hard limit why don they show 5 or 6 batteries in parallel? Is it just because they got lazy drawing batteries and stopped at 4 batteries ??

I am not sure though, hence asking here. Rather then form a false opinion and go forward on that opinion I formed its better to ask here as there are some very knowledgeable people in these parts.

 

[A.Justice]

In my opinion, Tesla is able to get away with paralleling massive amounts of cells be because of three reasons. 1, they're using legit matched and batched cells to start with; 2, they use cell level fusing, and 3, they have active cooling and thermal management on their packs.

I agree that you lose the ability to monitor anything at cell level after you start paralleling cells in a pack. I think that can be a calculated risk though, with matched cells and proper safety precautions, it will work just fine. There's TONS of UL listed stuff that uses paralleled 18650s, so there has to be a way to safely do it.

I actually have a 2p4s pack that I use regularly, I matched the cells as best as I could, and it's been about 2 years with no issues. Personally, I wouldn't go over 2p, unless they were really small cells. For a big LiFePO4 bank, I would rather make 2 or 3 separate packs, each with a BMS, that way I have more redundancy.

 

[Ampster]

is it a good idea to fuse LiFePo4 cells per each?

I have a 3P16S pack and have one Class T fuse. How many to parallel is really a function of use case, available space and need for redundancy. I have grid as backup and no need for redundancy.

 

[Ohms_Cousin]

I have a 3P16S pack and have one Class T fuse.

so what happens if a cell in one of 3P strings short circuits?

 

[Ampster]

so what happens if a cell in one of 3P strings short circuits?

It would probably drag dow the other two and the BMS would shut down the pack. I have never had that happen in ten years so it is a risk that I am not going to worry about. I did wire a Nissan Leaf module backwards once and it got warm which I corrected within a few minutes. Having a single pack with a single expensive BMS that I can monitor remotely has value to me.

 

[Ohms_Cousin]

In my opinion, Tesla is able to get away with paralleling massive amounts of cells be because of three reasons. 1, they're using legit matched and batched cells to start with; 2, they use cell level fusing, and 3, they have active cooling and thermal management on their packs.

I agree that you lose the ability to monitor anything at cell level after you start paralleling cells in a pack. I think that can be a calculated risk though, with matched cells and proper safety precautions, it will work just fine. There's TONS of UL listed stuff that uses paralleled 18650s, so there has to be a way to safely do it.

I actually have a 2p4s pack that I use regularly, I matched the cells as best as I could, and it's been about 2 years with no issues. Personally, I wouldn't go over 2p, unless they were really small cells. For a big LiFePO4 bank, I would rather make 2 or 3 separate packs, each with a BMS, that way I have more redundancy.

yes it makes sense to me that a 2P cell set is mundane enough to not worry about it further. But what if you get out to, say, a 6P string with LiFePo4 ? I may be over thinking this but this is on a boat - there is only one way out from below. Safety is paramount - at least to the point that the risk is acceptable.

Another way of doing this is to go 96v nominal. The electric motor I want to install can be 48v or 96v - can be ordered either voltage. At 96v it is 32S - so 2P32S = 4.5 kWh - so 11 batteries get me to my target capacity. There are advantages to 96v. The propulsion motor spins faster and so draws more air through the windings (its an air cooled brushed DC motor) - the manufacturer tells me at 96v it does run cooler. The motor draws half the amps at 96v Vs 48v (at 48v the projected power draw is approx 100 amps @4.5 kts speed). So drawing less amps from a higher voltage battery can only be a good thing for the cells I think? I mean at 96v the motor only pulls 50 amps - that has to be easier on the cells? I can get a high quality Australian made inverter for 120v nominal and it has an onboard charger so charging from the grid to a 96v battery is covered as well. (guessing it might be hard to find a 96v nominal charger powered by 230v from the grid)

I am not sure though - still researching all this

 

[Ohms_Cousin]

It would probably drag dow the other two and the BMS would shut down the pack. I have never had that happen in ten years so it is a risk that I am not going to worry about. I did wire a Nissan Leaf module backwards once and it got warm which I corrected within a few minutes. Having a single pack with a single expensive BMS that I can monitor remotely has value to me.

I thought you were looking into going to a JK BMS for each battery? Did you give up on that ?

 

[BradCagle]

You like to live on the edge huh.....

To me a 25P is a bridge too far. There have to be higher odds of a cell going bad on you when you times your chances by 25. At least it would seem so.

They were assembled this way from the factory. Cylindrical cells, just like in a battle born.

 

[Ampster]

I may over thinking this but this is on a boat

Redundancy is important on a boat and several small packs may be easier to service and place in various spots for good weight balance. If my installation was on a boat, I would go with redundant packs. Today good BMSs are reasonable in cost and Bluetooth monitoring is adequate.

 

[Ohms_Cousin]

They were assembled this way from the factory. Cylindrical cells, just like in a battle born.

yes the cylindrical cells seem to be tougher and more resilient

 

[Ohms_Cousin]

any really good reasons why 96v should be avoided? - other than shock risk, which can be managed.

 

[Ampster]

I thought you were looking into going to a JK BMS for each battery? Did you give up on that ?

No, I have had this $800 Orion BMS for four years and did not want get rid of it.