Mixing new cells paralleled with old

[sailingharry]

I am about to pull the trigger on EVE cells, about 300Ah each (not sure if 280 or 314 or something in between) for my sailboat. My current battery is 740Ah of AGM, and works wonderfully -- this would indicate that I should go with 8 cells (2P4S) for a significant step up in available power, especially if I spring for 314Ah cells (a 628 Ah bank with easily 500Ah available, compared to my existing bank with less than 400Ah available and challenges charging above 90%). BUT, I have this nagging thought that someday, I'd love to install AirCon (doesn't exist at present) and I could have a useful amount of run time if I hit evening at 80% SOC and am willing to run down to 15% SOC overnight. Yeah, that would be a brutal recharge the next day -- but it would be nice to have that option. To make that viable, I would really like a ~1000Ah bank (3P4S). But this is in the future. I have no need for it now, and LFP is going noplace but down.

Here's my thought. Buying 3P4S is an extra $400, and adds ZERO value at present. What if I buy 2P4S, and build my entire structure (box, bus bars, etc) for 3P4S and install 4 blocks of wood as the extra 4 cells. At some point, after I've installed AirCon and sussed out the viabilty of actually running on battery, I buy 4 new cells and they are a very simple drop-in to the existing bank (yea, balancing and all that -- nothing is "simple"). And, yes, in 4 years, there is a good chance that the new cells will be the same size (Eve likes this form factor) but they are likely to be larger (say, 350Ah).

Long held conventinial wisdom, dating back to flooded batteries, is that you can only parallel batteries of similar age, size, and capacity. I suppose that wisdom carries over to LFP. I've never bought that idea (but I've never actually mixed batteries either, so I don't have any experience). If the top and bottom of the charge curve (the knees) are the same, they should top balance out and hit 0 SOC together. Any cell that somehow gets there first will simply stop contributing -- it won't continue to go further (its not the same as in un-monitored series cells -- old NiMH for instnace -- where a cell can hit zero and be reverse charged and ruined. In my view, adding a third cell set to an existing 2 cell battery at some point in the future is certainly worse than tossing the old stuff and buying all new, but no worse and probably better than buying the full scale today and letting it age.

Has anyone added new cells to an old bank, and experienced negative outcome that you can share?

 

[Culchee]

I came across a forklift battery group on FB,and someone had the same question re paralleling batts of diff cap age etc ...I told them it wasn't a good idea...even though I hadnt
really done a lot of research myself..which I would normally do ...oh boy...they really came at me like a pack of piranhas..and rightly so..
A member from Aus kindly showed me an experiment that was carried out and proved conclusively that it didn't matter..I wish I could remember the process but the battery packs seem to handle being paralleled like old friends meeting new ones ..

 

[Culchee]

However...will lifepo4 behave in the same manner as lead acid..I've no idea...

 

[toms]

There are no issues adding old LiFePO4 in parallel with newer cells.

 

[wpns]

Cells or batteries? No problems paralleling new and old batteries, assuming the cell count and chemistry is the same), but I'm not sure I'd add parallel cells to an existing battery (or how to even do that without tearing it down and starting over).

The new and old batteries will charge and discharge at different rates, but you'll use 100% of the capacity of all of them as you go from fully charged to fully discharged.

 

[SomethingAppropriate]

You would have to size your wire to max for both batteries, on all batteries.

Internal resistance is different and both batteries could receive close to 100% current by themselves, while in parallel.

Also, you will find that after long or heavy charge or discharge, one battery will charge the other for a while even with no load or charge to balance themselves. This in turn adds extra and unneeded cycles to your new battery.

All things to consider.

 

[sailingharry]

I'm an engineer so I have a mind that ponders. Doesn't make me any smarter (especially since I'm a Mechanical Engineer), just predisposes me to think. It's a curse....

So here's my thought. At Float, all the cells are seeing float voltage. If the old ones have the same float voltage (we don't modify our settings over time, so I assume they don't change), then at float they all are at 100% charge (old ones may take a little longer, or something, but who cares). At (or near) dead low SOC, again, all are at 1% SOC (again, we don't change our BMS settings for cutout as the batteries age). So all cells go from "Full" to "Empty" and all cells put out all they have (all are rated 100Ah, old ones may put out 90, new ones may put out 105, but who cares if they all put out all they have?). In between, one or the other may put out more or less than they should -- but since they are banked together (each 3P set) they are continuously being "top balanced" every charge.

Now, what about anomalies. Say the old cells reach bottom faster. As soon as they start to hit their knee, they stop contributing and stop discharging (their internal voltage is microscopically less than the collective set voltage (as read on that groups bus bars) and they put out 0.00 Amps and the newer cell is carrying 100% of the load. Since my loads are typically less than .5C of a single cell (and typically less than .1C of a single cell!), it doesn't even harm that one. Same at the top end -- any cell that reaches the upper knee first just stops taking any more charge amps because its internal voltage is at the bus bar voltage.

One long standing comment that is often said is that new cell drain their current into old cell. An old cell will not discharge to a voltage less than the set of cells, so no cell can discharge into another cell (excepting an internal short, etc).

Now, lest anyone say, "If you know so much about this, why ask the question?" Because I'm fairly confident, but certainly not all-knowing. I've asked questions where I'm confident -- and come out of the discussion with a... well, a "new confidence." LOL.

 

[sailingharry]

I should reiterate that my bank will be built as if it were all 12 cells. 4 cells will be blocks of wood. In fact, the layout is such that it will be two on each side of the assembly, so the "blocks" don't need to be precision. The bus bars will be built to support the full bank. Wire sizes are based on my peak load (300A in theory, fused to 400A with appropriate cables). Upgrading will require me to remove all bus bars, loosen the compression, drop the new cells in place of the blocks, and reassemble. Top balance would need to be done to get close, followed with an appropriate time with the bank at float to allow the BMS to finish the top balance.

 

[Ohms_Cousin]

if one of your batteries is at a lower voltage to the other batteries, wont amps flow between batteries (from the high voltage battery to the low voltage battery) ? Say one battery is at 55 volts and one battery is at 50 volts - would there not be a big inrush of current to the lower voltage battery from the higher voltage battery?

 

[wpns]

if one of your batteries is at a lower voltage to the other batteries, wont amps flow between batteries (from the high voltage battery to the low voltage battery) ? Say one battery is at 55 volts and one battery is at 50 volts - would there not be a big inrush of current to the lower voltage battery from the higher voltage battery?

You are correct. You should match the voltages within {0.1? 0.25? 0.5?} volts before connecting them in parallel, and have OCP devices on each battery so if it sources (or sinks) too much current from the rest of the bank it'll trip a circuit breaker or blow a fuse. Technically, the BMS should disconnect the battery on over{charge, discharge} but you don't want to depend on that, turning off semiconductors doesn't always work...