My Project Lithium Battery Caught Fire

I missed so many interesting discussions.
The most interesting thing is that I predicted this. Proof.

[PROJECT] Lithium battery for Prius | PriusChat

 

1) In hybrid batteries that use lithium-ion cells, Toyota has balancing.
2) Tesla is an electric car, the battery mode is completely different. Hybrids (HEV) almost never (with rare exceptions) are charged to 100%
3) You are wrong

I guess the problem is not the cell chemistry, but the fact that he uses cheap cells that are not matched to the same capacity. This leads to overcharging of some modules and undercharging of others.
The root cause is low-quality (NOT automotive grade) cells.

 

Today I finally got around to reproducing the failure mode @sworzeh reported in post#1. @sworzeh was spot on with their "smells like Sharpie" statement... when these cells vent, their electrolyte reeks like a river of industrial strength uncapped sharpies stuffed down your nose!

...

I was unable to reproduce the failure on my first attempt, which was as follows:
-First, I charged a cell at 170 amps until it hit 3.65 volts, and then;
-I maintained 3.65 volts until the charge current dropped to 30 amps, and then;
-I then immediately switched to discharging into a 20 mOhm load, which is ~183 amps @ 3.65 volts, and ~140 amps @ 2.80 volts.
-Once the cell dropped to 2.80 volts, I immediately repeated the charge cycle.
-In total, I repeated this test four times back-to-back.

Results:
-The cell did not vent.
-The pack enclosure melted, but did not burn.
-The external module temperature was 80 degC when the test finished. It's likely the internal temperature was much higher.
Note: these test currents likely exceed NexCell's specifications
...

After this first test, I immediately switched to a neighboring cell and began test two:
-First, I charged a cell at 100 amps until it hit 3.65 volts, and then;
-I maintained 3.65 volts until the charge current dropped to 30 amps, and then;
-I then immediately switched to discharging into a 20 mOhm load, which is ~183 amps @ 3.65 volts, and ~140 amps @ 2.80 volts.
-Once the cell dropped to 2.80 volts, I charged the cell at 100 amps without a voltage limit, and then;
-When the cell hit 3.90 volts (3.65 volts is maximum 'safe voltage' for LiFePO4), the cell forcefully vented (enough to rapidly move a 40 pound concrete block resting on top of it). The vented electrolyte condensed on the surrounding concrete surfaces.

Results:
-The cell vented when overcharged. Note that the V1 NexCell circuit board is unable to alert the OEM BMS that a particular cell is overcharged, hence the V1 hardware cannot limit regen when a cell is overcharged.
-A thermal event did not occur.
-The cell under test did not appreciably expand.
-The cell under test was just over 115 degC, as measured by a thermal imager.
-Following this test, the cell was discharged to 0.0 volts and the terminals are now shorted to remove all residual energy.

...

Based on my results from the first two tests, I suspect that to reproduce @sworzeh's thermal event, the cell needs to get really hot prior to overcharging. This allows enough energy to accumulate inside the modules, such that they melt - rather than burn - until the cell vents due to internal pressure build up.

Note that since I was only testing QTY1 cell - and not the entire pack - I was only sinking a few hundred watts into a single cell... the neighboring cells were neither charged nor discharged. In the case where the entire pack is charging/discharging, there will be more heat build up for a given charge/discharge current.

...

Overall findings:
-overcharged NexCell cells vent without burning. This is expected due to LiFePO4 chemistry.
-overheated NexCell cells melt without burning.
-it's likely that some combination of both overcharging and overheating could cause a thermal event to occur.

 

Fortunately, I don't think there is such a thing. I'm pretty sure Toyota has the numbering system where there isn't a P and C (for example) with the same 4 follow-on digits, just like no DTCs have the same numbers for subcodes. With C1241 being a real code, there will be no P1241. If the C1241 has a 344 subcode available (just an example), no other DTC will have that subcode as a possibility.

 

I'm not sure I would assume too much about Toyota's deliberate commitment to the pattern you've seen.

For starters, they're not in control of all the code assignments. All U0, B0, C0, P0, and P2 codes are standardized by SAE, and those guys aren't doing anything fancier than assigning numbers in each range in order. There's no reason, say, a chassis code C051A (left rear wheel speed sensor - mechanical), and a powertrain code P051A (crankcase pressure sensor circuit), a network code U051A (invalid data received from door switch E) can't all exist.

You don't see these particular codes in a Prius because it hasn't got a mechanical left wheel sensor, or a crankcase pressure sensor, or a communicating door switch E. But if it had any of those things, Toyota would use the SAE code for it, rather than making up codes of their own in the manufacturer-defined ranges. In fact, they have changed manufacturer-defined codes they used in earlier Prius generations to be the SAE standard codes for the same things, in later generations after SAE standardized codes for them. The rarity of such overlaps between Prius codes probably comes down a lot more to coincidence than to design.

As for the INF codes, a lot of that may come down to history too. For example, in gen 1, there weren't many SAE standard codes for hybrid vehicle inverter issues, and Toyota made up the manufacturer-defined code P3125 to cover pretty much anything that could go wrong with an inverter, using a whole slew of different INF codes.

In later generations, after SAE defined a whole bunch of P0 codes covering different inverter problems, Toyota moved to using those instead, so now all the different things that can go wrong don't all have the same DTC with different INF codes—they're now parceled out into many different SAE standard DTCs—and they also still have the INF codes they had in the earlier generation. So it happens that the INF codes don't overlap, among all the various standard DTCs that all used to be P3125 with different INF codes. There are plenty of other examples where that happened.

That history tends to explain why it looks like INF codes are non-overlapping between DTCs, but as to whether Toyota ever sat down on purpose and said "we're never going to have overlapping INF codes between DTCs," I wouldn't count on it.

 

RIP Nexcell Lithium battery. Quality system is required for every industry as we saw how it changed Japanese industry in 1960's

 

Excellent write up.

While I understand the codes are used for demonstration purposes only, (and are noted as not being actual Prius codes), you still were "gotcha-ed" by the code description fortune cookie.

A "wheel speed sensor - mechanical" code would indicate that the ecu detected a problem with the signal that's caused by a mechanical fault (such as a damaged tone ring). It's not a circuit fault.



Posted via the PriusChat mobile app.

 

FWW, I never thought it was skills or installer tissue. I'm say its could be based on age of car and something like one harness terminal with just the right amount of no seeum corrosion versus the alternative on the battery/soother related. that's just IMO I also know nothing to save two lives about batteries, maybe one tho for sure.

 

Did you used signal shooter?

 

Is the car really that pink/magenta color as shown in the picture? If so, how did that happen?

 

For other than the OP, this may be the best question of the entire thread!!

 

I haven't seen much discussion about the install details of this particular pack:

What was the install / assembly process that was used to install this particular pack?

Did you reuse existing bus bars?
How corroded were they / what cleaning process was followed to remove corrosion?

Torque spec for the bus bar nuts?
New or existing bus bar nuts?

There are a few other nuts in the battery assembly that are considered "one time use" by Toyota. Were any of these replaced / properly torqued?

Photo of the Battery ECU terminals, were they corroded?

Torque spec for the bolts that screw into the bottom of each blade? (I believe these help ground the individual blades?)

Was the pack as a whole properly torqued with the 5 final bolts into the chassis of the car? Clean and free of debris for a clean grounding connection?

What was the climate where this pack operated? Was it climate control garaged or stored outside?

 

Torque spec for the bolts that screw into the bottom of each blade? (I believe these help ground the individual blades?

The bolts that fasten the individual modules to the battery case (bottom) are M5x0.8, the same as the module (+) and (-) studs. Torque spec is 48 inch-pounds. They DO NOT ground the module to the case, at least not intentionally. The steel insert on the module is knurled on the outer surface and pressed into the plastic module body. Sometimes, over the period of use, small cracks may form in the plastic, allowing electrolyte to make contact with the insert and sometimes leaking out around the "pressed in" insert. This is a frequent cause of a P0AA6 code, because it effectively allows the electrolyte to make electrical continuity with the battery case, which is in electrical continuity with the car body. You can see this very often as brown stains on the case floor under the ends of the modules

 

Your response was so informative, thank you for that.

Is the P0AA6 issue unique to the project lithium blades or also the factory design?

 

The P0AA6 code is provided to give an alert to any failure of isolation of the high voltage system, anywhere in the car. If there is any way for high voltage to find a conductive path to the car body or any other wiring in the car, P0AA6 will be shown. The car is checking for that all the time. There are also a few different INF codes to go with P0AA6 that help narrow the search down to a smaller part of the high-voltage system.



Battery modules developing cracks and oozing some electrolyte has always been one way that a P0AA6 code can crop up (with INF code 612), even with factory modules. But the code will be given for a high-voltage "leak" for any reason anywhere in the car.

 

That's one I have not heard of before. More than two cars and the excess went to the kids? When my parents split all I got was a move across town and a transfer to a different school district.