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

[slimf]

End of the day, have the batteries where you can afford to lose everything in the area & enjoy the cheap power while it lasts.

This

I built a shed and put my powerwall in it.. If it burns down, I'll cry because the shed cost 30k and the powerwall over 13K. But at the end of the day, it can burn down and not hurt anyone.

 

[ChrisD5710]

I notice that the bus bars are zip tied directly to the spacers. The plus bus is only a fraction away from the can of the cell. There is this very thin fragile wrapper only. If the bus bar gets hot and melts a bit into the plastic, the bus bar may short to the negative can of the cell, and NO fuse will save this.
Therefore, I have modified all my banks. I have raised the positive bus bar using a tie strapped underneath the bus bar, thus preventing the bus bar from ever getting near the cell.

Stay safe

ChrisD

 

[Minus1724]

from the looks of the orintation of the packs, where they are black ( positive and negative ) and the placement of the polarity of the packs. It could have been a shortcut between the busbars.

 

[Korishan]

I will have to agree with you guys. Looking at this pack:

you can see how close the bus bar is to the casing. That's gotta be less than a mm.

And if the fire started here, on this side:

it is very likely the bus bar made contact with the Neg shell.

@hbpowerwall perhaps you can do a measurement check between those bus bars and the shell/wrappers and maybe look for any that might appear to have been 'rubbed' getting thinner.

 

[Roland W]

I will have to agree with you guys. Looking at this pack:

you can see how close the bus bar is to the casing. That's gotta be less than a mm.

And if the fire started here, on this side:

it is very likely the bus bar made contact with the Neg shell

If the positive bus bar shorted to the cell neg, then the neg collector bus bar would as well need to be in contact with the shell to draw current uncontrolled from the rest of the string bypassing the 2nd fuse. That fuse did not blow as well. Two places of contact at the same cell opposite to each other is highly unlikely.
Such a 18650 cell has around 2Ah of energy stored in it. If if internally shorted and lets say produced a 20A internal current, that would make a 10C discharge in 6 minutes. Guess it might be to much for an old cells with unknown first life.

 

[Korishan]

If the positive bus bar shorted to the cell neg, then the neg collector bus bar would as well need to be in contact with the shell to draw current uncontrolled from the rest of the string bypassing the 2nd fuse. That fuse did not blow as well. Two places of contact at the same cell opposite to each other is highly unlikely.
Such a 18650 cell has around 2Ah of energy stored in it. If if internally shorted and lets say produced a 20A internal current, that would make a 10C discharge in 6 minutes. Guess it might be to much for an old cells with unknown first life.

This would only be if the short was a full on short. However, a slow short, one that barely makes contact, can drastically have a negative effect on the device(s) being involved in the short. There have been many houses burned down because wires in the wall had been frayed and were barely making contact slowly heating up until ignition. But yet, the breaker never sensed a large enough of a short to pop.
So if the bus bar has made a tiny contact with the shell and was constantly shorting around 1A, neither of the fuses would ever blow, but that cell would slowly heat up until catastrophic failure.

 

[completelycharged]

The one thing I have often wondered about is the whole issue of soldering to the -ve side of the cell with a large soldering iron creating a lot of heat into the cell. The cells all (appart from the odd one or two) seem to have good flux flow indicating good heat, but given Li cells don't like hot conditions and degrade very (relatively) quickly, prolonged soldering iron contact to me is an interesting situation. The +ve side of the cells tends to be spaced off away from the internal chemicals so less of an issue.

Spot welding creates high very localised temperatures with a low depth of heat dissipation, very unlike a large soldering iron held on the cell for a few seconds to get the flux to flow at upwards of 180C/360F (enough to boil electrolyte).

Maybe a complete non issue due to the lack of similar issues from DIY builds, just a thought and question that I have often wondered about.

Li cells almost seem to be like nuclear reactors in terms of thermal runaway.....

The safest and best environmental (temperature) location to me is to dig a hole a couple fo meters deep and put the cells in... temperature is stable yearr round and nothing to burn.

 

[Dr. Dickie]

80 temp sensors is easily doable. Especially with multiplexers and cheap thermisters.
There is always DS18B20 that use 1 wire i2C.

I know thanks this from HP I an going put temperature sensors everywhere.

 

[Dr. Dickie]

The one thing I have often wondered about is the whole issue of soldering to the -ve side of the cell with a large soldering iron creating a lot of heat into the cell. The cells all (appart from the odd one or two) seem to have good flux flow indicating good heat, but given Li cells don't like hot conditions and degrade very (relatively) quickly, prolonged soldering iron contact to me is an interesting situation. The +ve side of the cells tends to be spaced off away from the internal chemicals so less of an issue.

Spot welding creates high very localised temperatures with a low depth of heat dissipation, very unlike a large soldering iron held on the cell for a few seconds to get the flux to flow at upwards of 180C/360F (enough to boil electrolyte).

Maybe a complete non issue due to the lack of similar issues from DIY builds, just a thought and question that I have often wondered about.

Li cells almost seem to be like nuclear reactors in terms of thermal runaway.....

The safest and best environmental (temperature) location to me is to dig a hole a couple fo meters deep and put the cells in... temperature is stable yearr round and nothing to burn.

Soldering verses welding 18650s

And from daromer here:
Soldering 18650s

 

[hbpowerwall]

From experience, if the busbar touched the cell, fuses vanish instantly.. but it's a good point to add a spacer. I'm going to 'rebuild' the battery today to try make some more assumptions. At this point, my official opinion is that Poor design & maintenance is the issue more on this later.

 

[ChrisD5710]

The one thing I have often wondered about is the whole issue of soldering to the -ve side of the cell with a large soldering iron creating a lot of heat into the cell. The cells all (appart from the odd one or two) seem to have good flux flow indicating good heat, but given Li cells don't like hot conditions and degrade very (relatively) quickly, prolonged soldering iron contact to me is an interesting situation. The +ve side of the cells tends to be spaced off away from the internal chemicals so less of an issue.

Spot welding creates high very localised temperatures with a low depth of heat dissipation, very unlike a large soldering iron held on the cell for a few seconds to get the flux to flow at upwards of 180C/360F (enough to boil electrolyte).

Maybe a complete non issue due to the lack of similar issues from DIY builds, just a thought and question that I have often wondered about.

Li cells almost seem to be like nuclear reactors in terms of thermal runaway.....

The safest and best environmental (temperature) location to me is to dig a hole a couple fo meters deep and put the cells in... temperature is stable yearr round and nothing to burn.

You do not need a huge soldering iron that heats up the entire cell. My cells are fused with a 1 A Fast blow Fuse. They do not need much soldering, just enough to fasten the thin fuse lead. My solderjoints are seldom more than a few sq mm.

1 Amp per cell?? Sure 5600/14 is still 400 Amperes at 48 Volts or some 20 Kilowatts. I only utilize 2 x 5 KW inverters. (0.5 A per cell at full load)

Immediately after removing the soldering iron, the call can be touched with a finger without being burned.

I know, covering the entire cell bottom with solder using a huge iron is not healthy for Your cells.

ChrisD

 

[hbpowerwall]

This soldering on his batteries is 100x better than my own.. That said I think I've found the issue. I can say that poor maintenance & improper design contributed to this near miss. Not much hotter, but they don't get better from this point only worse..

 

[nz_lifer]

This soldering on his batteries is 100x better than my own.. That said I think I've found the issue. I can say that poor maintenance & improper design contributed to this near miss. Not much hotter, but they don't get better from this point only worse..

Looking at how much that heat is radiating, it answers my question of how many temperature sensors we would need to pick up on rouge cells.
Minimum of one per cell pack, more if you wanted to catch it early.
Then it's a case of many low accuracy ones or a few high accuracy ones.

 

[hbpowerwall]

that's a 5deg difference.. you would need temp sensors on every cell in random order so hotter cells stand out. if you had temp sensors
1 2 3 4
5 6 7 8
9 10 11 12 etc

and the hot cell was say "7", 2 3 4, 6 8 & 10 11 12 would also be slightly warmer so you might see the difference but it wouldn't stand out. making the numbering of the temp sensors random would make 7 stick out like a sore thumb and the rest less so.

Also on the temp difference one cell I just checked was 0.2v less than what it was yesterday being 3.8v now if that cell was 4.2 v that number 7 cell would be MUCH hotter. Also if you had two cells heating beside each other (this is what I suspect was the cause ultimately) those two cells would accumulate heat much faster heading towards thermal runaway.

 

[OffGridInTheCity]

In regard to heat. Is there any info on the load (amps/cell) that was placed on this battery? Do you know if it was 1a or 2a/cell (more stress/heat) or light as in 200ma/cell (less stress)? And the range - was it routinely charged up high as in >4.1v and/or discharged low as in <3.0v?

 

[Bubba]

that's a 5deg difference.. you would need temp sensors on every cell in random order so hotter cells stand out. if you had temp sensors
1 2 3 4
5 6 7 8
9 10 11 12 etc

and the hot cell was say "7", 2 3 4, 6 8 & 10 11 12 would also be slightly warmer so you might see the difference but it wouldn't stand out. making the numbering of the temp sensors random would make 7 stick out like a sore thumb and the rest less so.

Also on the temp difference one cell I just checked was 0.2v less than what it was yesterday being 3.8v now if that cell was 4.2 v that number 7 cell would be MUCH hotter. Also if you had two cells heating beside each other (this is what I suspect was the cause ultimately) those two cells would accumulate heat much faster heading towards thermal runaway.

I wouldn't randomize the sensors because it would make debugging more difficult.
Could you better see a heater by using Mean compared to max deviation displayed highlighting the bad cell (or something like that)?

 

[hbpowerwall]

No load info at this point, BMS info was - 4.08 - 2.5v, shunt trip was misconfigured so not working as was the canbus

 

[hbpowerwall]

I wouldn't randomize the sensors because it would make debugging more difficult.
Could you better see a heater by using Mean compared to max deviation displayed highlighting the bad cell (or something like that)?

That was a brain fart - made sense to me.

 

[daromer]

There is a very simple solution to the issues regarding the safety around cells...
1. Dont use lithium ion cells
2. Dont have them in any of your houses. Ie have them in a shed 50meter or more away from your home in proper container
3. Buyt them prebuilt.

Using 2nd hand lithium ion cells will never be 100% safe thats a fact. Since not even new LiIon cells are 100% safe if you look at how many scooters or mobiles blown up during the years.

With that said we can make it safer. And its not about taking it to the extreme and think that cell level fuses save you from a cell-blow out because it will NEVER do that. Cell level fuse only saves from dead shorts. And i would say that 9/10 fires was not near a dead short. This also based on alot of reading from datasheets and tests including talking to testers/builders of the cells.

One thing is to limit when shit hits the fan. Keep the cells outside. keep them stored in such way that they cant light each other and so forth.
Next is to monitor and try to act before it happens.. This is tricky because there arent much data out there on how except of keeping track of temperature
What we know is that above 80C of the cells they become highly unstable and at 120C they can be unstopable.

Having alot of temp sensors is one way of doing it but there is 2 other ways i have done work with myself and with manufactures of high current systems that i can share

1. Ir video system that monitors each pack graphically and looks at if any part of the image is hotter than others. This though takes time.
2. Forced air tunnels where you meassure temperature in and out. Doing this you can easily sense if the temperature raises just slightly and can act. You need to have a baseline first doing this.

If there is interest in having any of the 2 above in the DIY world let me know and i might be able to show some type setups of this.

(Yes there is more ways looking at it but for mixed packs it gets more complicated)

 

[nz_lifer]

2. Forced air tunnels where you meassure temperature in and out. Doing this you can easily sense if the temperature raises just slightly and can act. You need to have a baseline first doing this.

This gets my brain going
After some time in operation you would get a baseline of the temperature delta of a normal day.
In - Out vs ambient. Thor humidity in there too?

But depending on volume being measured the rise from 1 bad cell may not be measurable at bank level.
Maybe at pack level?