They were charging without a BMS. I think the moral of this is don't be an idiot.
I understand. If I give you a box which says "BMS" on the top, it solves the problem. If you don't buy the box, you are an idiot. Is this proper way of engineering such safety-critical systems?
The sad part is, the more serious an incident is (such as this), the
less likely people seem to be willing to do a proper root cause analysis - or even
speculate properly. For some reason, people get scared, and our brains seem to have this failure mode of fearing analysis when exactly that would be your best friend - instead we resort to easy simplifications and finding clear scapegoats - having no actual information doesn't prevent you from faking it. Typical suspects are, depending on "which bubble you live in", that a li-ion fire is either caused by a BMS, or by the lack of it. When properly analyzed,
either can be the actual reason, but most of the reasons are neither of these two.
Speculation is all good, to find ideas, but in cases like this, some real information is needed, and we should be honest about it when we are speculating.
Here comes some speculation:
Most typical reasons for catastrophic failures include improper manufacturing quality and physical damage (such as cell penetration). This is said to be a dismantled Tesla pack, so it should rule out cell quality (they would be Panasonic, one of the best you can get). This in fact makes the incident really interesting - the root cause is probably some of the less frequent stories.
After these two obvious causes, the next on the line might be using cells that have died by severe overdischarge. As nctnico explains, copper dendrites may appear. There are so-called proper "revive" algorithms for slightly overdischarged cells (basically, very slow charge such as C/100 with voltage monitoring - AFAIK the idea is to have the dissolved copper go back in the current collector where it came from, instead of forming said dendrites!) While officially supported by the manufacturers such as Samsung, it may still be better to just dispose of such cells, even if they are almost within limits.
A proper BMS does not allow charging of a pack with a cell at, say, less than 2.5V. On the other hand, there are improper BMS ICs on the market from reputable manufacturers, which fail to follow the correct practices. For example, a certain TI product does the "revive conditioning" at C/10 when configured typically, and it's impractical to configure it below C/20. The conditioning charge is too fast, and the dendrite risk is real. People who blindly design in these trusted BMS products, are doomed. But like with any semiconductors (think about an opamp!), you need to understand the datasheet, all its parameters, and you need to understand the problem you are solving; no opamp datasheet guarantees it's suitable for a specific application. Now you would expect that application-specific ICs are suitable for their specific applications, but don't take it for granted; these ICs are not safety-certified finished products;
the BMS designer designing in the chip is the one who is supposed to be in that position, but often they don't know this!
So, as you can see, the next speculation target might be "the failed BMS" story. Yes, you read correctly, this is the case where either the BMS is to blame directly, or the misuse of a BMS, coupled with the false sense of security in using one. When people charge or discharge
without a BMS, they
usuallyl tend to have a clue, set a charger CV voltage setpoint properly, for example. With a BMS, a lot of utterly dangerous advice exists, even from the manufacturers. The most fundamental issue is the "just use BMS and forget about it" cargo cult attitude, as demonstrated here, again.
As a concrete example, many BMS manuals and brochures instruct that with their BMS, you can use "any" charger, even "stupid" current sources, because the BMS "controls" the charging. In an ideal model this works, as long as:
1) the BMS never malfunctions - when wired like this, it stops being a redundant safety system, and becomes the only critical loop, and it tends to be a more complex, more error-prone system, compared to a simple resistor divider limit in a power supply;
2) the BMS is actually connected properly, so it
can physically stop the charging - preferably two ways, soft "charge enable" and driving the main contactor. No joke, this mistake really happens. It doesn't help if an unconnected "charger disable" signal screams while the cells start approaching 5 volts.
A proper use of BMS should
definitely include using a charger with a properly adjusted voltage setpoint, so even if the BMS fails, the charger just charges the cells fully as a single pack (like you do without a BMS). Even with a slightly imbalanced pack, this is
unlikely to cause dangerously serious damage, so it acts as an extra safety layer. Independent safety layers are great! That's where an independent BMS system can really shine. Don't ever waste this opportunity by trusting it completely as the only mean!
With recycled cells, you never know the full story of the cells. For example, if these packs are dismantled from a crashed car, mechanical damage is well possible. Or, a previous owner has charged a dead cell, causing those dendrites, and the current owner cannot know this as the cell isn't below low voltage limit anymore.
Or, it may be a combination, for example, too high of a charging current to a deep-discharged cell, combined with a physical shock, combined with it being charged fully.
Speaking of dendrites, lithium plating and lithium dendrites do happen near full charge, by:
1) overcharge by voltage - this is easy to define and easy to avoid.
2) charging current too high, combined with high state-of-charge, combined with low temperature. This is a more complex function.
This is a different mechanism from overdischarging, which causes copper current collector to dissolve, letting copper ions into the electrolyte.
Home
Help
Search
Login
Register
