Finding & dealing with a "bad" cell in a 2P8S battery

[balachai]

I don't have a "bad" cell yet, just trying to understand what I'm signing up for with a "parallel first" 2P8S configuration as the cells age. Eventually one or more cells will develop problems. Is my understanding correct?

(a) If a cell develops an internal short-circuit (say a slow short with low current) then its partner parallel cell will tend to discharge into it, pushing current backwards through the bad cell and heating it up. The good cell's voltage will drop as it discharges. The BMS should identify low voltage in this cell pair. During balancing, this cell pair will take longer to raise up to the correct voltage. If the cell pair's low voltage is severe then the BMS would switch off the entire battery.

(b) Alternatively, internal connections could break within a cell raising its internal resistance. When the battery is being charged, its partner parallel cell will accept a greater current than the "bad" cell and get temporarily overcharged to a high voltage. Meanwhile the "bad" cell receives less current and gets undercharged. Once charging stops I guess this overcharged "good" cell will slowly discharge into its undercharged partner balancing the cell pair again. Eventually, repeated temporary overcharging would cause some kind of failure in the "good" cell - once the entire pair is "bad" I guess its high or low voltage will be picked up by the BMS and the entire battery would be disconnected.

(c) A cell could fail completely "open-circuit" with no current flowing through it at all - it's effectively removed from the battery. During charging and discharging, its "good" partner parallel cell would experience twice the current of other cells in the battery. The BMS is now monitoring the "good" cell alone in this cell pair. Its voltage would vary much more than other cells, prompting the BMS to disconnect the battery.

If I'm managing a battery with problems a, b or c, I can identify the problem cell-pair from the BMS info, either watching in real time during battery charging or from a data log. But how to find out which cell in the pair is "bad"? Bypass one cell* with an appropriate-length heavy-duty cable, see what happens over 12 hours, then do the same with the other cell? After finding the "bad" cell, how do I deal with it?

LiFePo4 -- Parallel or Not? - Page 2 - Cruisers & Sailing Forums

Originally Posted by mbartosch And this is a very important one. One of the things I have learned over time is that it may look very easy on paper, but it may be difficult to execute in limited and

www.cruisersforum.com

 

[snoobler]

(a) a failed cell in parallel with others will destroy the others as well due to discharge to 0V.

(b) concur

(c) concur

At the first detection of an anomaly where you suspect a failed cell, break the cells into 2X 4S batteries and use a BMS on one while letting the other sit. If you've installed the BMS on the good one, you won't see any issues and the one sitting will likely exhibit voltage drift where one identifies itself as an outlier.

The question becomes, why not just spend a little extra $ and have monitoring at the individual cell level?

 

[balachai]

break the cells into 2X 4S batteries and use a BMS on one while letting the other sit.

Makes sense, but since I'll have 16 cells, in my case that would be 2X 8S, right? Is it practical to do this while leaving the cells in place?

The question becomes, why not just spend a little extra $ and have monitoring at the individual cell level?

I'd love to have individual cell monitoring and don't mind spending a bit more for it. The issue is for my voltage (24V) and required capacity (>10 kWh), I'd need two cell strings in parallel, 8S2P. The downsides for me:
- I'm scared of eddy currents after reading the document by Orion BMS.
- I'd like to use CAN bus to signal from BMS to Victron MPPTs & inverter, but I can't figure out how to achieve if I have parallel strings, each with their own BMS.

http://orionbms.com/manuals/pdf/parallel_strings.pdf

 

[snoobler]

Yep. Sorry. 8S.

Why the fear on the eddy currents? The presence of a BMS on each string addresses their concerns.

Anyway, yeah, CAN bus interface complicates it and drives you to a single 2P8S configuration.

Another option is a Batrium. You could have 2X parallel 8S batteries all monitored by a single bms and 16 blockmons with Victron compatible CAN interface. Spendy though.

 

[balachai]

Another option is a Batrium. You could have 2X parallel 8S batteries all monitored by a single bms and 16 blockmons with Victron compatible CAN interface. Spendy though.

Cool I'll check it out, thanks!

Why the fear on the eddy currents? The presence of a BMS on each string addresses their concerns.

Well I'm just a noob and they're BMS experts. So I pay attention when they say things like
"A single weak or bad cell will lower the voltage of the string it is in and cause other paralleled strings to force charge the string with the bad cell. This not only can lead to a dangerous situation..."

Orion says that a BMS and contactor on each string fails to address a lot of problems (see below). On the other hand I understand that a lot of people use this approach without problems, including Steve_S. So I'm unsure what to think.


Orion's view on parallel strings each with a BMS & contactor:
Problems this approach addresses:
- Protects cells
- Works with most equipment

Problems this approach does not address:
- Eddy currents (only somewhat controlled)
- Lower usable capacity
- Inrush currents when strings first paralleled
- Lower maximum power
- Possibility of cascading shutdown
- Exponential impact of one bad cell
- Individual control of charge and discharge currents
- Possible damage from interrupted charge power (from cascading shutdown)

 

[Steve_S]

A BMS can ONLY see what is at the end of the wire lead. It has no way to know if there are paralleled cells, it simply sees the paralleled set as one unit. If a cell in the set goes bad, it can affect the rest of the set & downstream if it's really bad.

A Battery Assembly such as straight 8S for 24V, then each cell can be monitored & managed by an 8S BMS. Parallel cells into that, the BMS still only sees 8 cells. You can't drop a 16S BMS in there - just won't work.

Paralleling cells is playing the "Averaging" game.
Bulk/Commodity / "Unmatched" cells are NOT recommended for paralleling. Too much differential/deviation potential.
Matched/Batched & Binned cells can be used in Parallel because they are tested to match Voltage, Internal Resistance & Impedance through their entire working range.

EV's use Parallel Cells BUT they are all as closely matched & batched as possible, otherwise, the packs would fail prematurely and even then some do anyways. It is even more critical for EV's as they are using other Chemistries which CAN explode & catch fire, unlike LFP. That is also another reason why EV Batteries are so darn expensive because the tolerances have to be bang on and consistent across a lot of cells.

 

[jtvt]

Yep. Sorry. 8S.

Why the fear on the eddy currents? The presence of a BMS on each string addresses their concerns.

Anyway, yeah, CAN bus interface complicates it and drives you to a single 2P8S configuration.

Another option is a Batrium. You could have 2X parallel 8S batteries all monitored by a single bms and 16 blockmons with Victron compatible CAN interface. Spendy though.

IIRC Batrium says there is no point in having more than one block mon on a parallel cell since each cell in parallel will show the same voltage. Parallel batteries is ok. There doesn't seem to be a solution for parallel cells.

 

[balachai]

Yeah, no point having multiple BMS leads on cells paralleled, I get that. I think snoobler was talking about having a single Batrium BMS connect to all 16 cells in a 8S2P configuration i.e. two separate 24V strings. That looks possible:

In that case would you want/need a contactor for each string, or just one for the entire battery?

 

[jtvt]

I'm thinking that the multiple string set up with the batrium could only use a single contactor. AFAIK, The batrium doesn't have the ability to trigger a relay based on the state of some of the cells (Say cells 9-16 (bank#2) vs cells 1-8(bank#1)).

I don't know whether the batrium can keep multiple strings balanced. However, the batrium should be able to balance single cells better than paralleled cells.

Seems to me the main problem with multiple strings is they become out of balance.

 

[balachai]

In that case would you want/need a contactor for each string, or just one for the entire battery?

Looks like one contactor for the battery, and a fuse for each string, like in the second schematic here:

Batrium BMS setup 8s3p

I'm building a new battery bank with 28 EVE 280Ah cells, and I'm wondering wether the following setup is valid. At 0.5C the battery pack can continuously handle about 400A, but in reality I will only use 200A (0.25C). Everything is designed on 400amps though. The system is 24V. The Batrium...

diysolarforum.com

 

[jtvt]

I'm not sure what the point of the fuse is? I'm guessing to avoid massive current moving from one bank to another if a cell dies thereby avoiding one of the issues Orion points out.

I've been trying figure out whether to go parallel cells vs. parallel batteries. I have the same questions as you. In my case, I'm doing 32 cells for 48 volt and if that works I'll add another 32 cells. I was previously leaning toward parallel cells, but now I'm on the fence again. I already have the Batrium. I'm not using a contactor, but a shunt trip breaker.

I'm wondering about the scenario where you have two battery strings monitored by the Batrium and if the Batrium can balance the two given that the two will not be exactly equivalent. There will be differences in resistance, etc so there will likely be one bank that (dis)charges faster than the other. Will the Batrium be perpetually trying to balance one bank? Is this really a problem? I don't know.

IIRC, I did read that it was possible to use multiple mons on one cell thereby increasing the ability to balance cells. I'll have to see if I can find that doc.