My Project Lithium Battery Caught Fire

By anyone who is not blind, delusional, biased, senile or has had one too many glasses of hateraid ol chap.

 

"YOu DOn'T knOw whO I Am, BUT bUY tHeSE DoZEn eXpeNsIVe dOODADS to Get INcreASed MPg. USE my LInk tO SAve sOme CASh."
"Okay, I bought one and it caught on fire."
"LIAR!!! Tell Us WHo YoU aRE!!!"

 

See my proposed action plan in post #223 in your Signal Soother Test thread.

 

@Mr. F

Follow the troll train the link to my thread where I told old chapman who I have met from the forum.

Otherwise Im not sure what kind of personal info you want to know about me or why it matters to you.

Honestly it seems like you have mental issues typing like a keyboard warrior who would never meet anyone in the first place.

 

Seriously people, get a life. My NiMh battery caught fire FFS, I posted photos of it sometime back on this forum, yet I still had some members posting it was false
I have tortured a lot of LFP cells when I was initially designing our BMS and battery systems for off grid house batteries, and they simply do not catch fire .... they vent their more volatile parts of their electrolyte, it might look like smoke, but it is actually a dense cloud of vapour that settles very quickly, but they can't actually catch fire, they do not generate their own oxygen, so any fire would require oxygen from the air outside the cell casing and something to ignite it.

It requires cobalt in the chemical make up to generate oxygen inside the cell when over heated, then you have all the requirements for a battery fire, LFP, LYP and lithium titanate do not contain cobalt, Sodium ion doesn't contain cobalt either, so no chance they will catch fire either.

Sodium ion is a cell chemistry still developing and has been over 3 to 4 yrs, there are over 700 different combinations of anode, cathode, electrolyte mixes and more still in the lab.

Even though popular social media experts seem to consider the electrolyte is salt water, that was abandoned in the very early testing, the water breaks down and shortens the cycle life, the non active plate (anode and cathode relate to the direction of electron flow, charging or discharging) is not graphite, hard carbon and graphene are popular at the moment, but CATL is already on its second commercial version, so it's way too early to say what the ultimate combination will be .... but you can bet it will come out of China ... they are just so far ahead of the game in battery technology now the rest of the world would be some what foolish to even attempt to play catch up ......
Will sodium ion be the chemistry type to replace lithium ion ..... who knows, it will be the one that can charge the fastest, discharge at a consistent current rate from 100% SOC to 5% SOC without serious degradation of capacity or internal resistance over 10,000 cycles (that equates to 3 cycles from 100% to 5% every day for 10 yrs) yet remain cost effective .... sodium ion is up there with the best of them at the moment ......

T1 Terry

 

@T1 Terry
You sound like an intelligent person, which is refreshing. Thank you for taking the time to respond to my accusations.
Looking forward to continuing this conversation with another rationale adult.

A key difference is that you designed a BMS into your LFP system, whereas NexPower did not design a BMS into their LFP packs. So then all it takes is one cell overcharging and then it's game over. They added insult to injury by later creating the "Signal Soother", which masks cell failures to the one remaining OEM component that might otherwise detect said failure.

Would your company allow you to sell an LFP product that only monitors stack voltage every ten cells (e.g. 1-, 10+, 20+, 30+, etc)? Do you agree that is a recipe for disaster?

Please search youtube for several counterexamples. While fires are less common with LFP cells, they can and do happen, both directly & indirectly (e.g. due to heat spread into nearby combustibles). I have a few addenda below.

Yes, you have described the most common LFP failure mode. However, while LFP batteries are vastly less fire-prone than other lithium chemistries (e.g. NMC), they're not completely immune to fire.

I agree. Specifically, LiCoO2 is the primary catalyst for self-sustained oxygen generation in 'standard' lithium cells. The same precursor also decomposes into carbon monoxide and lithium oxide (the white crystalline solid that gets deposited on nearby surfaces during thermal runaway).

However, I must clarify that while LFP cells aren't self-sustaining during a fire, they are still capable of burning (i.e. producing flames during a thermal event). Yes, non-LFP lithium is more likely to run away, but the fact remains that LFP cells can burn if not properly monitored. Please consult your references online for numerous examples.

Taking a step back, do we really expect customers to care whether or not flames are emitted during an LFP thermal event? Either way they're going to have a bad experience, which is easily preventable by adding a BMS. Obviously having the car 'only' fill with smoke is a better outcome than also having it also burn to the ground... but are we really ok with a product that doesn't make even a passing attempt at preventing the thermal event from occurring?

Whether or not the cells sustain fire during a thermal event, engineering best practices require LFP systems to have a properly designed BMS system. That's why you designed a BMS for your system, right? The fact is NexPower chose not to add a BMS to their LFP products, which is unsafe because it leads to thermal events (whether or not there are flames).

I agree, and will reiterate that only a few sodium chemistries will ultimately 'win' in the long run... time will tell who picked the right horse. Five years from now we'll likely have a similar conversation regarding the "NMC vs "MnO" debate ten years ago.

Yes, I agree, and time will tell. Two additional deliverable are volumetric & gravimetric energy density... sodium has a long way to go there... if it can catch up, it will leave lithium in the dust.

 

No sure how up-to-date you are on the latest sodium ion developments, but testing shows the charge faster with less heat generated and discharge much faster with less heat generated and very little voltage drop ..... but it is a tad more linear than LFP, so that needs to be factored in. Their real advantage is the apparent zero damage to being discharged to 0V, I'm not willing to attempt reverse charging them yet, where a cell drops to 0V but those either side still hold voltage and capacity, so the current is dragged through the 0V cell effectively reverse charging it .... fatal for LFP cells, I'll wait until I finish torture testing these cells before I subject them to that though, just in case :lol:

I've had LYP cells held at 4.5v till all the electrolyte had boiled off (they won't go over that voltage while they can boil off their electrolyte) but after that, the voltage is meaningless, there is nothing linking the plates together, the heat becomes so high the plastic separator sheet melts and seals each plate.

Care to link an LFP fire video that wasn't caused by aluminium cell cases touching one another or an outside metal component linked to the battery negative?
Unfortunately, I no longer have the time laps video of a cell on charge with a gas flame mounted 50mm above the vent ... it actually blew the flame out, second test was 150mm above the vent and the vapour stream can clearly be seen below the flame and a very weak flame above it, still not enough oxygen mixed with the vapour to create a decent flame. At $200 per cell, two was enough to sacrifice ....

You mention a vehicle filling with smoke, that indicates combustion, so not the correct term I believe, it is vapour and smells so sickly sweet that no one could stay seated in the vehicle well before it built up to cloud stage ...... from that one would assume the event of the vehicle filling with a cloud of vapour would not happen.

As far as batch testing cells, you do realise the Toyota NiMh module system in the genuine battery only voltage tests in batches of 12 cells (6 cells per module and 2 module testing) but I agree, every cell should be voltage monitored and a charging/discharging cut triggered when the upper and lower extremes are met .... but not even Tesla monitor that closely .... MG does, but I see they aren't part of the US market ......

Let's look at a battery pack that monitors 4 cells in each group, take a cell above 4v or below 1v and the safe range for the other 3 cells plus the over voltage or under voltage cell adds up to a voltage that could be monitored ... not that I consider that to be an optimal arrangement, but enough to avoid thermal runaway .... so safety level monitoring, not, not optimal cycle life monitoring .....

I consider that proper cell voltage monitoring, whether a single cell or multiple cells in parallel is the optimal method and temperature monitoring is not required, but if series group monitoring is used, them temp sensing is a must. The next essential thing is proper cell balancing, not milli amps but 5 amps min that switches on at 150mv out of balance and off at 5mv, so the unit isn't wasting battery capacity nit picking that last bit or trying to balance unnecessarily because different internal resistances were causing the voltage differential, not actually capacity related ... a hard yet not impossible task for Prius 228v battery, 16 groups x 4 cells with an average voltage of 3.56 volts. The nom. voltage for a traction pack is 201v/16 groups/4 cells = 3.1v per cell, the sodium ion can handle 4v high to almost 0v low, 3 x 4v = 12v x 16 = 192v if every group lost a cell yet it attempted to fully charge the battery ..... it wouldn't be that hard to set a min and max voltage regarding charging and discharging ..... better to monitor every cell, but .....

T1 Terry

 

@T1 Terry: Once again, thank you for taking the time to have a civil conversation.
@AzusaPrius: Look how we both take considerable time and effort to answer each other's statements.

I have seen lab data reporting the same.
However, I will hold off on further conjecture until I've reviewed NexPower's (allegedly proprietary) specific sodium cell chemistry in their V3 pack. Until then, let's continue discussing their existing V1.x & V2.x LFP designs.

IMO, sodium is more than a "tad more linear than LFP". In fact, LFP is known for it's nearly flat midband discharge curve, whereas sodium remains linear throughout the entire discharge range. For reference:
Typical LFP discharge profile.
Typical Na-ion discharge profile.

Using the above representative graphs, we can see the following rough voltages (I'm only using one decimal place for simplicity):
Vcell@25%SoC (LFP): 3.4 volts
Vcell@75%SoC (LFP): 3.5 volts
delta: 0.1 volts

Vcell@25%SoC (Na-ion): 2.4 volts
Vcell@75%SoC (Na-ion): 3.8 volts
delta: 1.4 volts

So then the sodium cell has a 14x larger voltage delta over the middle half of the SoC range.

I agree 0 volt tolerance is a huge benefit to sodium ion cells.
However, I'm more interested in what happens when these sodium cells are overcharged. If V3 doesn't have a per-cell BMS (very likely), that's the more dangerous condition; a cell at 0 volts has basically no energy inside, whereas an overcharged cell has all it can store... this is the reason pushing more energy causes rapid unscheduled disassembly.

Rest assured, I intend to test these V3 sodium cells on the bench.

Clarification needed: Are you actually referring to Lithium Iron Yttrium Phosphate cells (LYP), or did you mean to write 'LFP'?
When I tested V1 NexPower LFP cells to failure, they failed well below 4.5 volts.

It's more boring, but here are some more scientific academic research papers that performed real-world tests:
https://scijournals.onlinelibrary.wiley.com/doi/full/10.1002/ese3.283
Thermal runaway and fire behaviors of lithium iron phosphate battery induced by over heating - ScienceDirect

Less academic, but more approachable sources:
Can LiFePO4 Batteries Catch Fire? Unveiling the Science Behind the Flame

And for those that just want to watch videos:

(this is a puncture test, but I reference one observation below)

Note that half of the off-gassing from an LFP cell is hydrogen gas, which is of course explosive. So then when you contain this gas inside a vehicle, the interior space is highly susceptive to ignition. One potential ignition source is the overheating module itself, which we know can occur with NexCell's V2.x product (see OP's pictures in this thread).

...

If the above sources aren't authoritative enough, let's take a step back and ask ourselves "does a commercial LFP battery require a BMS"? I think you'd be hard pressed to find a reputable engineering team that would sign on shipping an LFP battery product without a BMS that can disable the system if even a single cell voltage is out-of-bounds.

I hesitate to ask the hypothetical question "name one commercial LFP product that's shipped without a BMS", as someone would likely state "NexPower's V1.x & V2.x Prius Batteries do! Checkmate!"

Note that in an enclosed space (e.g. the vehicle interior), the ignition source doesn't need to be in the outflow jet stream.
Were you testing for ignition in an enclosed space, which gives the 50% hydrogen gas outflow time to mix with atmospheric oxygen?

Based on the pictures @sworzeh's posted, we can probably both agree that a layperson would call the particulate matter inside their cabin "smoke".

This semantic differentiation isn't particularly relevant to the layperson. However, I agree that we should strive to use the correct terminology in our discussions. In general, I prefer the term "thermal event", because it's much more inclusive to any possible condition where things overheat (e.g. fire, smoke, meltdown, burns, etc).

FYI: Both @sworzeh, myself, and my wife described the smell as "strong permanent marker". I did not smell anything 'sweet' in the NexPower V1 cells I tested to failure on my front porch.

I agree, and they should do everything they can to leave the vehicle immediately. However, consider the hypothetical where someone drives the car to a state where a cell is about to fail, and then parks the car in their garage and goes inside... minutes later the cell fails, filling the car with explosive hydrogen gas. Meanwhile, the cell is rapidly heating, and can easily achieve temperatures that would ignite properly mixed hydrogen gas.

This is likely what occurred in the third video I linked above: When the person stabbed the cell with their pick, they briefly inserted enough oxygen into the cell that the hydrogen inside the cell was then able to ignite due to proximal contact with the red hot internal cell structure. This same condition could easily occur over time in a sealed enclosure.

Absolutely. The OEM Honda NiMH systems are the same, too. This is acceptable with NiMH systems, due to their vastly different chemical reaction properties. I will go into details if I must, but I would think you could spend 20 minutes exploring the interwebs. Do let me know if you want to debate this further.

This is the most important point, and I'm glad we agree on it.
Given that NexPower doesn't monitor every LFP cell voltage, and doesn't have a charge/discharge inhibit method...
...do you agree with me that their LFP designs are unsafe?

Maybe I'm misunderstanding your statement, but Tesla certainly monitors every intermediate stack voltage on every product they sell. For example, a 124S24P pack containing QTY2976 cells doesn't need QTY2976 'BMS' measurements, but does require at least QTY124 'BMS' measurements. A properly designed system (with high current interconnects between cells at the same voltage) only requires one cell voltage monitoring circuit per 'layer' in the stack. Note that for safety reasons you'll typically have two separate measurements per 'layer', but that's outside the point of this discussion.

Note that within a given parallel cell group, no cell in that set can exceed the single voltage measured unless it physically disconnects from the pack. This is the reason why Tesla has per-cell fuse elements built into their battery tabs. If a single cell goes bad and wants to operate at a different voltage, its fuse will open and it will disconnect from its neighbors. Once disconnected, the cell is no longer allowed to charge, which drastically reduces the risk that it will vent.

Other manufacturers don't have visible per-cell fusing in their welded tab matrix, but instead opt to use internally fused cells. They achieve the same result, albeit with slightly less efficiency.

We agree on this:
-A properly designed BMS must monitor each cell voltage level in an LFP stack (e.g. QTY48 circuits for a QTY48S pack).

I disagree.
Temperature monitoring is required on any LFP BMS, regardless of the cell voltage monitoring method.

We agree on this concept.
In addition, based on my testing, NexPower's V1.x & V2.x circuitry does not contain "proper cell balancing".

I disagree that high current balancing is absolutely required on a hybrid traction battery. You can achieve excellent pack balancing with even a low balancing current. Most lithium modules I've examined (including Tesla) use a relatively low current balancing circuit... typically this circuit is a simple passive discharge resistor, which can sink maybe 50 to 100 mA. I do the same on my LiBCM product, which can balance the pack at up to 52 mA. As long as your BMS system can remain on when the car is off, there's no reason to quickly balance a pack over its useful lifetime.

However, it wouldn't hurt to build an active stack-shuttling flyback converter that would shift energy from each cell, into a flyback transformer, and then back into the entire stack. LT (now Analog) makes some cool ICs that can cost effectively achieve these goals. In general it appears auto manufacturers shy away from this additional cost and complexity, as it only becomes worth the effort once the pack degrades to a state where the user should consider replacing the pack entirely.

Certainly there are mission critical applications wherein these active energy balancing designs make sense. Certainly I'd want redundant power sources, each with active balancing in a battery operated SIL3 and/or SIL4 product... but not in a SIL2 electric/hybrid vehicle application. It's overkill and adds unnecessary cost. I'm sure at least one car manufacturer has implemented active balancing, but I haven't seen it and certainly most companies don't.

Please clarify.
I didn't understand the point you were trying to make here.

Again, I'm holding off on sodium analysis until I get the V3 hardware.