Destructive battery pack test suggestions

[RagingGrandpa]

Planning for what to do next time I have a working but untrustworthy EUC pack... which could be a nice opportunity to confirm a few assumptions about EUC failures.

Please presume it will be a 24s2p 21700 pack, such as below.

(I also have some non-trustworthy but still functional Gotway controllers and motors, to consume in the experiment...)

My first ideas were to answer:

1. "How much battery current flows when a board is overloaded and burns?"
   (Helps validate fuse selection.)
   Wire a shorting path into a few FETs; apply the short and measure battery current.
    
2. "Do cells burn in the presence of a dead short outside the pack?"
   (Helps demonstrate/refute the need for interrupters inside the pack.)
   Apply a short to the pack output cables; leave pack unwrapped and use video to demonstrate if/where the circuit burns open.
    
3. "How effective is a garden hose at stopping a pack fire?"
   Wrap up the pack in shrink, begin at 100% SOC, initiate runaway using a heating element near one cell, wait until a few cells vent, turn off the heater, then try blasting the pack with the anemic household hose.

Other suggestions?

Thx

[Eucner]

Fascinating. Divide idea 3 to sub-merged and not sub-merged. I think even a small amount water can take a lot of heat away when sub-merged.

[Tawpie]

dang. we need to get you about 6 go pros so you have enough video angles. This sounds like FUN!

[Chriull]

10 hours ago, RagingGrandpa said:

Other suggestions?

Ad 2) could be interesting if in case of a short some components/PCB track act as fuse...

The other test can imho be performed without working BMS, too

Personally i'd be most interested to see what happens in the typical GW ?before EX.N? setup - battery directly above the controller - just separated by some thin plastic strip, motor wires very near/touching the big capacitors - with constant burden/overload. With slowy increasing motor current...

... But no idea how to perform such a test...

A bit more boring and time consumig could be continous charge/discharge cycle with single cell voltage logging. With different parameters - no balancing/balancing (80 vs 100% charging), discharge to ~3-3.15V vs ~3.5V under load. Maybe one gets one nice degradation pattern shown?

[Chriull]

13 hours ago, RagingGrandpa said:

"How effective is a garden hose at stopping a pack fire?"
Wrap up the pack in shrink, begin at 100% SOC, initiate runaway using a heating element near one cell, wait until a few cells vent, turn off the heater, then try blasting the pack with the anemic household hose.

Just saw again some li ion cell tortures - at 6:00 in 

one sees li ion cells on heaters.

So be prepared for fast flying glowing cells - could be hard to target them with a hose

Stay safe!

[Paul A]

Google search of lithium battery fire shows some results that might be worth considering if proceeding with such an experiment is worth the risks.

Have not fact checked the results or the credibility of sources.

There also seems to be contradictory information.

 

What happens when you put water on a lithium battery fire?

Hence, we should avoid attempting to extinguish lithium battery fires with water. The primary reason is lithium burns in the air we breathe. And moreover reacts with water to form explosive hydrogen that may develop a chain reaction.

 

 

Are lithium battery fumes toxic?

Lithium-ion batteries can produce dozens of dangerous gases when overheated, according to a new study from the Institute of NBC Defence and Tsinghua University in China. ... The gases, which are potentially fatal, can cause strong irritation to the skin, eyes and nasal passages, and harm the wider environment.

 

What happens if a lithium battery explodes?

A lithium-ion battery explosion can result in severe burns and, in rare cases, even death. Lithium-ion battery explosions can be caused by manufacturing defects in the original battery or by contact with metal objects that cause an external short-circuit.

 

Can lithium ion batteries burn without oxygen?

So the answer is; yes, lithium ion cells contain their own oxygen. You cannot smother a cell fire, you just have to get its temperature down.

 

How hot does a lithium battery burn?

The burning lithium creates a metal fire existing at temperatures of 2,000 degrees Celsius/3632 degrees Fahrenheit

[RagingGrandpa]

Experiments will be conducted with preparation and care, of course

[Paul A]

Ok, please take care.

[Chriull]

1 hour ago, Paul A said:

What happens when you put water on a lithium battery fire?

Hence, we should avoid attempting to extinguish lithium battery fires with water. The primary reason is lithium burns in the air we breathe. And moreover reacts with water to form explosive hydrogen that may develop a chain reaction.

Correct. It's seen in my above linked yt video how an opened harmless, because empty battery created some ?500°C/F? with some water poured on it!

But still the big advantage is that by water cooling intact cells do not get heated up and start a thermal runaway by the venting cells.

These venting cells burn anyhow - additional water won't drasticly change the happening.

1 hour ago, Paul A said:

What happens if a lithium battery explodes?

Technicly there is no explosion.  (if the safety vent/"predetermined breaking point" functions ...)

It's just hot gaseous burning electrolyt venting with quite high speed out of the cell...

..but still i'd prefer to not be too near to such an event

[mike_bike_kite]

8 hours ago, Chriull said:

It's just hot gaseous burning electrolyt venting with quite high speed out of the cell...

I'm pretty sure that hot gaseous burning at high speed inside a sealed container is just a description of a bomb.

[Tawpie]

1 hour ago, mike_bike_kite said:

hot gaseous burning at high speed inside a sealed container is just a description of a bomb

and once it gets a hole in it, it's a rocket! or if a piece comes off you're left with a bullet casing.

[RagingGrandpa]

A test article has arrived!

Backstory:

• EUC model: Nikola 100V 1800wh, purchased Sept 2020
• Cell model: LG 21700 M50T
• Pack removed from EUC by owner because of failure to charge beyond 98V, after 2 months of use. (Smart man caught it early!) Pack was replaced thru vendor warranty.
• Abnormalities found:
  Cell group 11 at 0V and internally shorted (0.2ohm DC resistance when disconnected from BMS).
  Slight rusting of weld-strip at cell group 11.
  Traces of powdery corrosion residue within pack. No evidence of liquids.
• Balancing circuit for group 11 is operating normally: with the dead cells disconnected, if I apply 3.6V to group 11, parasitic current is 5.5uA. I observed the same for group 12. And if I increase the voltage to 4.9V, 40mA flows due to balance resistor activation.

I plan to add a shorting wire across cell 11, to ensure the pack can sustain normal discharge currents... then proceed with destructive testing as noted above.

Edited December 18, 2021 by RagingGrandpa
Added balancing current observations

[supercurio]

Is there some more analysis you could could complete on the health of each cell group and root cause of the BMS failure @RagingGrandpa?

I'm asking because it could be invaluable if this pack(s) holds the key of how they tend to how/why they typically self ignite.

 

[RagingGrandpa]

The other 23 groups were healthy... but fine, I'll poke around more.

I don't think there's much news to be had though. If cells go to 0V, for internal reasons or external reasons, the results are well understood.
 

Edit:
Updated post above: BMS is working normally.
In general, it is possible for a stuck-on balance resistor to overdischarge a cell group slowly over time. But, no such fault is present in this BMS today.
I was curious about it because of odd balance-circuit resistances that I observed when the cell group 11 was detached from the BMS; but the values became normal (matching the good cell groups) once I applied a normal-range voltage (e.g. 3.6V) in place of the dead cells.

5.5uA parasitic load seems excellent! A full pack would take 200 years to self-discharge

Edited December 18, 2021 by RagingGrandpa

[supercurio]

Thanks @RagingGrandpa, I read the updated message.

So what could have killed the cell group 11?

Is the only valid theory that the pack manufacturer put 1 or 2 defective cells in there?

[RagingGrandpa]

Well, the cells were working "normally" when the EUC was new. Charging >100V with a working BMS means the cells must have been balanced (otherwise the BMS disables charging). And the remaining cells are still very well balanced.

So it leaves us with a cell-internal problem: some slow but substantial self-discharge occurred. Probably in just one cell, taking its parallel brother down with it.

It begs the question: "What could a pack builder do to identify and reject a bad cell like this one?"

• Per-cell capacity testing: probably not enough difference from good cells.
• Per-cell resistance testing: perhaps, yes.

... but if the self-discharge wasn't present until 5 deep cycles, it's nearly impossible for a builder to find.

 

Edited December 20, 2021 by RagingGrandpa

[supercurio]

Yes I was thinking the same. The quality and health of at least one cell must have been pretty bad compared to spec, and it died quickly.

Totally making this up but would it be reasonable to imagine these 3 class of pack builders exist, namely:

1. run a full-range capacity test (like 4.2V to 2.7V, 1C or faster) of each cell before building a pack, rejecting cells off-spec.
2. test the internal resistance of each cell, rejecting outliers
3. use every cell provided by the supplier, no test

And Begode pack builder being definitely in the 3rd category since the beginning?

Which was just fine when using Sanyo GA 18650 cells, fantastically consistent in quality.
And now leads to disastrous results in the offending 900 Wh packs.

[goatman]

im new to euc's but not to  batteries and why they put lg m50t cells in euc's is beyond me

an EUC is basically automatically and continuously going from discharge to charge as you ride, that act will kill the lifecycle of most cells rather quickly.

the only cells ive found while testing that actually like to be continuously cycled with no breaks without prematurely degrading was the 30q and the 40T.

thats what acceleration and braking is

alot of LG cells will say right in the datasheet to let the battery rest atleast 30 minutes or an hour before charging or it will kill the batteries cycle life.

I was very surprised and curious to see m50's in EUCs, they didnt even make my list of cells to test

are you people having good cycle life out of your packs?

 

[supercurio]

1 hour ago, goatman said:

im new to euc's but not to  batteries and why they put lg m50t cells in euc's is beyond me

an EUC is basically automatically and continuously going from discharge to charge as you ride, that act will kill the lifecycle of most cells rather quickly.

the only cells ive found while testing that actually like to be continuously cycled with no breaks without prematurely degrading was the 30q and the 40T.

I'd like to request you share your cell testing story, maybe in its own topic. I'll read attentively.

1 hour ago, goatman said:

thats what acceleration and braking is

alot of LG cells will say right in the datasheet to let the battery rest atleast 30 minutes or an hour before charging or it will kill the batteries cycle life.

This delay seemed surprising as for a few classes of devices using Li-ion batteries, a waiting time between charge and discharge is either impractical or even impossible.

I think what you refer to is the exact method the cell manufacturer describe in order to replicate their capacity testing, where they'll list voltages, charge and discharge currents, charging termination and the amount of minutes between each sequence.

But that's only to define the test spec I think, not to set requirements on how to use the cells themselves in production.

1 hour ago, goatman said:

I was very surprised and curious to see m50's in EUCs, they didnt even make my list of cells to test

are you people having good cycle life out of your packs?

I can't answer for everyone but the general sentiment from EUC users is that degradation is not an issue if even noticeable or even measurable in terms of real-world range.

Since the packs on EUCs are on the large side compared to other PEVs, they tend to be less cycled which probably helps.

[goatman]

in this thread from another forum

https://endless-sphere.com/forums/viewtopic.php?f=14&t=108547#p1587964

 

you will see this

#82 d-2410, c-2373
#83 d-2407, c-2369
#84 d-2406, c-2367
#85 d-2404, c-2366
#86 d-2404, c-2367 removed wait times between discharge/charge/discharge
#87 d-2402, c-2364
#88 d-2400, c-2364
#89 d-2401, c-2362
#90 d-2399, c-2361
#91 d-2398, c-2359
#92 d-2398, c-2360
#93 d-2398, c-2359 going back to 5min/2min wait
#94 d-2396, c-2357
#95 d-2397, c-2358
#96 d-2397, c-2356
#97 d-2395, c-2356
#98 d-2395, c-2355
#99 d-2394, c-2357

 

40T same thing, 40T3 is a really good battery for these wheels, less volt sag would probably give the same capacity if not more than the m50T's

5c discharges from 4.2 to 2.5 1000x, no problem

[RagingGrandpa]

1 hour ago, goatman said:

5c discharges from 4.2 to 2.5 1000x, no problem

Ok but we don't need a 5c discharge... for us a 1c discharge is extremely uncommon. (How often do riders drain their pack from 100% to 0% in a 60 minute ride, with a fullsize EUC?!) 

6 hours ago, goatman said:

are you people having good cycle life out of your packs?

Yes, we really do have some 400-cycle EUC's out there (20,000 mile equivalent on an MSuper). But more often users upgrade before 200 cycles, for various reasons not related to cell aging.

[goatman]

i do different types of tests that would simulate a  typical ebike ride using 10,7.5,5and 2.5amp discharge during the discharge cycle and then a 2amp charge per cell.

also for temperatures put a 40T and P42a in a freezer at -10c for 24hrs than do a 10amp discharge and youll see the p42a outperform the 40t but 40t and 30q can be "desulfated" every 100 cycles to recover capacity, same with the p42a and vtc6, not really desulfate, were not dealing with lead acids but i think it has something more to do with the sei layer.

i dont believe in 5c tests nor in those 1c lifecycle tests, i power traction vehicles with my batteries, 50E and M50s are storage cells, they both kind of suck and why are they still using Battery Murder Systems instead of active balancers? why arent the controllers going into thermal rollback instead of cutting out? maybe a mod can put this into a different thread???

its really not the discharge that would kill the batteries but more the switching to regen. for the average rider not a big deal but if your riding high performance wheels where youre braking hard from high speeds you go from pulling high amps to pushing high amps back into the pack. that would be destructive testing but if 400 cycles is 20,000 miles it will be hard to spot,except for the odd pack with a b grade cell in it, basically at 10,000 miles or 250 cycles you will start seeing the 50e packs start failing if not sooner at 10amps max/cell.

i dont know enough about euc's yet, i came here because they look pretty simple to mod or even make but when i went to go buy a control board from ewheels a banner says the wharehouse burned down and no stock, speedy feet has some in stock

what wheels and batteries are the ones that are failing or causing fires.

[mike_bike_kite]

13 hours ago, goatman said:

why are they still using Battery Murder Systems instead of active balancers? why arent the controllers going into thermal rollback instead of cutting out?

Can you pretend we don't know what an active balancer is and explain. Also why the standard BMS isn't so good. If the active system is better then can it be added to existing packs? cheaply? You'd get a gold star from me if you can explain why the Gotway/Begode wheels are suffering so much more than other makes of wheels. Keep in mind it's mostly affecting 100v wheels. Other manufacturers do use the same batteries. Also some wheels have caught fire before even being unboxed. Could you also mention your qualifications, if any, in this area.

[supercurio]

47 minutes ago, mike_bike_kite said:

Can you pretend we don't know what an active balancer is and explain. Also why the standard BMS isn't so good. If the active system is better then can it be added to existing packs? cheaply? You'd get a gold star from me if you can explain why the Gotway/Begode wheels are suffering so much more than other makes of wheels. Keep in mind it's mostly affecting 100v wheels. Other manufacturers do use the same batteries. Also some wheels have caught fire before even being unboxed. Could you also mention your qualifications, if any, in this area.

In short, active vs passive balancers

• active can take energy from a [high] cell group and charge another [low] cell group
• active can balance to 1A, up to 3A on some models whereas passive are limited to low currents dissipated in small resistors
• active can run cooler

Negatives of active balancers:

• often mis-configured in terms of voltage range, trying to balance before the top voltage and ending up un-balancing the pack instead.

IMO, passive balancer seem to be fine for top-balanced EUC packs, but need carefully crafted, smarter logic to handle repetitive 80%/partial charging use cases.

[goatman]

just a hobbyist, ill get into bms/balancer later

from the pictures above, firstly the nickel strip looks like its made for 18650 not 21700, maybe the op can measure the center on center

and can he say how the bms balance lead is connected, spotwelded or soldered, where the pen is pointing

 

 

just from a look at pictures above these are poorly constructed packs, no structure, looks like the spotwelds and shrink keep it together

the cell holders are the insulators, my money would be on they might be dead shorting at the positive end and the outer neg ring

 

 

 

 

Edited December 26, 2021 by goatman
trying to upload pictures?