When things go wrong... it can go really wrong!

hbpowerwall

Administrator
Joined
Oct 7, 2016
Messages
2,218
about 15 hours ago I release info that I have possession of a DIYPowerall that started to smoke and eventually catch fire on youtube

All I know from a quick 1-2 min on the phone was that a powerwall had caught fire & I was welcome to come pick it up.. Full of trust and fear I jumped in the van and went on my way..

Few hours later I was back at home pondering my entire project and if I could sell it all and buy a Tesla Powerwall. A two min roadside conversation took place where he cranked out a whole lot of info while shoveling cells from one van to another like drug dealers trying not to look suspicious

The short story is - Fire alarm sounded, the battery shed was found to be full of smoke, cells were cut out of their wall mount, rushed outside & at some time while carrying them out fire started shooting out from two of the cells. Hose was aimed at all the batteries laying on the grass and they had a good spash of water to cool them down - with two 40P batteries ended up inside a bucket of water for a few hours.


I've taken a bunch of photos of what I have here check them out, its scary stuff...
IMG_0723.JPG

IMG_0736.JPG IMG_0735.JPG IMG_0734.JPG

IMG_0733.JPG IMG_0732.JPG IMG_0731.JPG IMG_0730.JPG IMG_0729.JPG IMG_0728.JPGIMG_0727.JPG IMG_0726.JPG IMG_0725.JPG
 
Last edited:
Heat would have conducted through the busbars and just melted the solder.
Ah ok. Well in that case, heres another theory. Could it be possible that, with the packs being stored on there sides, a heater cell can get so hot that it melts the solder which in turn runs down the positive end cap onto the negative wall and causes a dead short. This would explain the fuses still being intact and if it's been a heater for a while, could have compromised the CID. If it's cheap Chinese solder, the melting point could be within reach of a heaters temperature range? Dunno, just spit balling as I'm as worried as you are. Possible? Plausible? Dunno ;-)
 
Ah ok. Well in that case, heres another theory. Could it be possible that, with the packs being stored on there sides, a heater cell can get so hot that it melts the solder which in turn runs down the positive end cap onto the negative wall and causes a dead short. This would explain the fuses still being intact and if it's been a heater for a while, could have compromised the CID. If it's cheap Chinese solder, the melting point could be within reach of a heaters temperature range? Dunno, just spit balling as I'm as worried as you are. Possible? Plausible? Dunno ;-)
That is exactly what the supect cell in the video looked like, solder ran into the ridge between pos and neg.
 
I'm inclined to agree with a few others who have mentioned an internal short (on the YT comments), aka a dendrite growth, that caused an internal short. Because they were almost fully charged when it happened means the fuse would of made no difference anyways as there wouldn't have been enough current flow to pop them. It had enough internal energy to self ignite.
This is definitely a reason to have multiple thermal sensors though. As I'm sure that the cell didn't got hot "just this once". It would of been something over time getting hotter and hotter, thereby weakening the cell structure.

Maybe this is something that Jaron can incorporate into his longmons to have several temp sensor ports. Or perhaps an addon monitor (tempmon(?)) that would connect in series with the longmons and allow deeper monitoring of the packs.
 
When I watched the video, the first thing that came to my mind was it probably was an internal short. As @Korishan suggested I think it may have been dendrite growth, maybe due to charging at below 0°C?
 
Last edited:
Heat would have conducted through the busbars and just melted the solder.
Make sure those cells disconnected from the busbar are also fully discharged.

Are there voltage graphs/logs for the hours~days before the incident that indicate that the problem cells/pack was self-discharging? Wondering if this happened suddenly without warning, or indication of a developing problem was missed.


[...] maybe due to charging below 0°C?
They're in Brisbane Australia, so it's mid summer now. And most Aussies have never seen snow their entire life :)
 
They're in Brisbane Australia, so it's mid summer now. And most Aussies have never seen snow their entire life :)
I know that they're in Australia, The dendrites could have grown half a year ago, and the internal short could have "started" due to vibration or heat. But when there are no temperatures below or near 0°C ever, this falls out as a possibility.
 
I'm struck by the rusted pack/cells and the comment that it 'was in a bucket of water' for a while - and therefore that caused the rust. However, in my experience things don't rust in 24 or 48hrs... is it possible the rusted cells were rusted over time somehow / exposed to water vapor or condensation or ? and this lead to the failure?

*I could be wrong about rust in 48hrs and/or the amount of time they were in the bucket of water - so I'm just thinking out loud here*
 
I'm struck by the rusted pack/cells and the comment that it 'was in a bucket of water' for a while - and therefore that caused the rust. However, in my experience things don't rust in 24 or 48hrs... is it possible the rusted cells were rusted over time somehow / exposed to water vapor or condensation or ? and this lead to the failure?

*I could be wrong about rust in 48hrs and/or the amount of time they were in the bucket of water - so I'm just thinking out loud here*
I have seen metal that was heated and quickly put in water rust in a very short time. It is curtanly possible with the presence of dissimilar metals.
 
I'm struck by the rusted pack/cells and the comment that it 'was in a bucket of water' for a while - and therefore that caused the rust. However, in my experience things don't rust in 24 or 48hrs... is it possible the rusted cells were rusted over time somehow / exposed to water vapor or condensation or ? and this lead to the failure?

*I could be wrong about rust in 48hrs and/or the amount of time they were in the bucket of water - so I'm just thinking out loud here*
Have you ever looked at your brake rotors first thing in the morning?? They are rusted quite a bit. The heat generated would of been enough to melt the nickel plating and cause rusting within hours.
The other option is that the guy who assembled them used a dremel, or other abrasive pad, on the ends to make the "solder stick better". Then once the heat melted the solder off, they'd be exposed.
 
Are there voltage graphs/logs for the hours~days before the incident that indicate that the problem cells/pack was self-discharging? Wondering if this happened suddenly without warning, or indication of a developing problem was missed.

I think logging/monitoring is key to preventing these catastrophic failures.
Got to pick it up problems as early as possible.
 
Rust forms within minutes/hour if exposed. Nothing strange there. :)
 
soooo need to get a fire resistant outdoor thingy for batteries

....

This is the reason I haven't put my wall up. I have 4000 cells here tested, cataloged, barcoded... but haven't built the packs...
 
Heaps of replys... but #1 theory is definitely dendrite growth, looking at all other batteries they are all built well, I mean better than mine.. Looks like thought has gone into the build especially if you consider the age of the build.

End of the day, have the batteries where you can afford to lose everything in the area & enjoy the cheap power while it lasts.
 
>#1 theory is definitely dendrite growth,
Going with this then... what do we do about it in terms of risk assessment. Obviously not all cells catch fire. For example, I have Fujitsu LifeBook with 18650 cells that are now over 15yrs old, charged/discharged many years in a row, still sitting around in my office. There are *many* examples of older laptops and their packs sitting around and you don't hear about fire from this.

I think I'm asking for more 'insight' on this. Do dendrites form at random, due to abuse only, due to age? What about old laptops with 18650 - are these more or less risky to have around than DIY powerwalls formed with them? Won't Tesla cars/powerwalls (larger cell counts than laptops) be a risk as they age as well?
 
Last edited:
I'm struck by the rusted pack/cells and the comment that it 'was in a bucket of water' for a while - and therefore that caused the rust. However, in my experience things don't rust in 24 or 48hrs... is it possible the rusted cells were rusted over time somehow / exposed to water vapor or condensation or ?

Note that the "rust" is only in the small gap area between the positive plate and the negative sidewall. I think that was caused by the voltage and whatever minerals was in the tap water. What's the proper word: Electrolysis? Electroplating? Galvanization? Galvanic corrosion?
 
>#1 theory is definitely dendrite growth,
Going with this then... what do we do about it in terms of risk assessment. Obviously not all cells catch fire. For example, I have Fujitsu LifeBook with 18650 cells that are now over 15yrs old, charged/discharged many years in a row, still sitting around in my office. There are *many* examples of older laptops and their packs sitting around and you don't hear about fire from this.

I think it's important to note all of the safeties in commercially-made batteries that we remove when we are recycling the cells. Temperature sensors are a huge point here. Some laptop batteries even have 2-3 sensors in one pack of 6 cells. When these batteries detect an issue, the battery is disabled permanently in some cases. That's why we're able to open them and find good usable cells. No BMS is going to be able to manage this in a battery pack of 80 cells. You would need 80 temperature sensors...
 
80 temp sensors is easily doable. Especially with multiplexers and cheap thermisters.
There is always DS18B20 that use 1 wire i2C.
 
Last edited:
Back
Top