Shunt snapped and caused a fire🔥

[k490]

Due to the fact that DC can maintain an arc across a larger gap once it gets started is it possible the shunt opened up under high load and an arc continued to burn until the wood caught fire? I know this can be a problem with solar panel disconnects where open it up under high loads the breaker arcs and starts a fire.

 

[OffGridForGood]

Due to the fact that DC can maintain an arc across a larger gap once it gets started is it possible the shunt opened up under high load and an arc continued to burn until the wood caught fire? I know this can be a problem with solar panel disconnects where open it up under high loads the breaker arcs and starts a fire.

I wondered about this as well:
Looking closely at the shunt, the plates have small cuts in them, I can imagine under high enough current, there could be arcing across this gap quickly creating heat. If the shunt bar were to crack across, but remain close enough for the arc to persist, the breakers may not trip, and the arc under high amperage conditions quickly heats up to ignition temperatures. As @Hedges pointed out, one breaker appears to have been heated to the point it melted, and perhaps couldn't trip.
If all 93 PV panels were connected to the system, there could be 1,000A (51.2v) available to charge the packs. OP noted 300A shunts, and from another thread, issues with breakers shutting off. We can imagine 1,000A across 10 packs = 100A each pack,{OP says highest current 60A, so perhaps less that 93PV panels installed} but imagine a breaker trips, then a second one, a thrid; Now 1,000 Amps is charging six packs, then five, then four with the amperage per pack rapidly increasing.

Eventually something fails, either a crimp, or the shunt - as the OP says in post #1 - or a BMS fails to cut off charging at high current or high cell voltage.
Luckily smoke was noticed, emergency crews were called and responded, property damage was limited, no loss of life. So many of us now have an opportunity to review and learn something.

OP has batrium - should be able to see the individual cell voltages: this would show if any cells were dropping out of charging - ie if a breaker did trip, and if any cells hit over current or over voltage - ie if a BMS failed to protect cells.

 

[42OhmsPA]

@nezjnr how are things going with the cleanup?
When do you plan to start the next build?

 

[Checkthisout]

Good catch on the corroded shelf support.

 

[Mattb4]

My inspection is complete and my power back on. One issue noted was the billion dead carcasses of Asian beetles (looks like lady bugs) discovered in the main disconnect/meter panel at the pole. Some mud dauber nests on the inside of the panel also. Darn pests! Anything outside at my place with a hole is soon cemented over. The battle with nature is an ongoing one.

 

[robbob2112]

I wondered about this as well:
Looking closely at the shunt, the plates have small cuts in them, I can imagine under high enough current, there could be arcing across this gap quickly creating heat. If the shunt bar were to crack across, but remain close enough for the arc to persist, the breakers may not trip, and the arc under high amperage conditions quickly heats up to ignition temperatures. As @Hedges pointed out, one breaker appears to have been heated to the point it melted, and perhaps couldn't trip.
If all 93 PV panels were connected to the system, there could be 1,000A (51.2v) available to charge the packs. OP noted 300A shunts, and from another thread, issues with breakers shutting off. We can imagine 1,000A across 10 packs = 100A each pack,{OP says highest current 60A, so perhaps less that 93PV panels installed} but imagine a breaker trips, then a second one, a thrid; Now 1,000 Amps is charging six packs, then five, then four with the amperage per pack rapidly increasing.

Eventually something fails, either a crimp, or the shunt - as the OP says in post #1 - or a BMS fails to cut off charging at high current or high cell voltage.
Luckily smoke was noticed, emergency crews were called and responded, property damage was limited, no loss of life. So many of us now have an opportunity to review and learn something.

OP has batrium - should be able to see the individual cell voltages: this would show if any cells were dropping out of charging - ie if a breaker did trip, and if any cells hit over current or over voltage - ie if a BMS failed to protect cells.

Shunts are made by embedding the conductive strips in the end pieces, either by casting around the strips or solder/braze the strips to the ends. Then they make cuts in them to calibrate the resistance to the correct value. For the soldered ones they can reheat and move the end slightly. The strips are a copper alloy called 'manganin'. On the better shunts the ends are solid blocks of copper, on the cheaper ones they are another metal.

 

[HarryN]

Thanks for sharing the info. It helps us all.

At one time I was toying with the idea of putting batteries in a box surrounded by baking soda or similar. When it gets hot, it releases CO2.

Didn't ever do it, but this thread reminded me to think about it again.

Harry

 

[Checkthisout]

So is everyone essentially agreeing the crappy shelf collapsed?

FYI, I absolutely hate those shelves. The particle board is only 1/2 and they pull apart when gently slid or tugged on wrong.

OP's install actually isn't otherwise too bad. Would just get better racking in future.

Would also suggest putting the whole setup in a small shipping container outside instead of in garage/house.

 

[robbob2112]

Thanks for sharing the info. It helps us all.

At one time I was toying with the idea of putting batteries in a box surrounded by baking soda or similar. When it gets hot, it releases CO2.

Didn't ever do it, but this thread reminded me to think about it again.

Harry

Baking soda around generic metal will conduct if damp and also eat into the metal. It is great for neutralizing acid, but not so great for the purpose you list.

 

[robbob2112]

So is everyone essentially agreeing the crappy shelf collapsed?

FYI, I absolutely hate those shelves. The particle board is only 1/2 and they pull apart when gently slid or tugged on wrong.

OP's install actually isn't otherwise too bad. Would just get better racking in future.

Would also suggest putting the whole setup in a small shipping container outside instead of in garage/house.

Just my revised opinions -

A. The battery pack in question overcharged/overheated and potentially didn't trip the breaker due to the one negative wire that looks like it could have been shorted to the frame of the shelf. Once the battery overheated and bulged it may have ignited the shelf board from the contact heat where it was sitting. then havock ensues.

B . The shelf sagged or gave way enough to tug some wires out of place or cut the insulation. Given the batteries are nearly from one edge to the other with a small space between I doubt the partical board actually gave way totally. Assuming the shelf metal was connected to the battery negative (we don't know this) all it would take would be a positive to short to it from before the first breaker which didn't trip. and fire, and other havock ensued.

The shunt connected to the negative breaking would have just caused the battery in question to drop out. If there was already a max current going it _may_ have caused the others to take to much charge current before they tripped breakers. I choose to think the shunt broke when the shelves fell and jerked wires verse being the cause of things.

The salt in the dry extenguisher would have also caused the breakers on the unburned battery packs to trip so they may not have been involved at all.

An to sum it up, we will never know what actually happend until the OP comes back and shares the rest of the tale of woe and more pictures with us.

 

[Checkthisout]

Somebody may have already said this but the shelf upright could have gotten cooked when current shorted to it.

Ha ha.

Anything over 10kwh of storage needs a black box to record failures.

 

[k490]

Does anyone make a smoke detector that can shut down the inverter?

 

[hwy17]

So is everyone essentially agreeing the crappy shelf collapsed?

I'm not convinced of a collapse. I have some of these shelves stacked with oil and they sag but they haven't collapsed.

The batteries lacked separators and were constrained with exposed threaded rod.

Perhaps sagging allowed for cell movement causing some kind of internal short in the bank, putting incredible heat into the main terminal wiring by thermal conduction and either melted the shunt solder or melted the wire insulation, leading to a more spectacular short. Or an internal short caused a vent flame, and the fire collapsed the rack and that caused more shorts.

I'm not opposed to the collapse first theory, but I see other possibilities.

 

[Supervstech]

It is unlikely we will ever truly know what caused the failure, there are so many potential failure points.

I see corrosion either from conditions, or arcing. It is possible the shelves necame energized, and arcing ate the supports away, generating heat that ignited the shelves, and the fire just snowballed the issues.

 

[Checkthisout]

Already pointed out but just to elaborate. If shelf gave way on bracket that's collaped.

The weight of the battery pack would pull those positive leads down onto the now angled shelf end bracket and would easily cut the insulation. Those wires are pre-breaker and I assume go directly to the end of the packs so full juice without fusing.



However, are those breakers doing the positive and negative or just two positives? If that's a positive and negative in the same breaker, then I say with 100% certainty that it was that shelf end bracket that gave way and allowed the pack to drop down.

So in the above photo, it's clear those two leads shorted and the insulation burned off because the wire got hot vs the leads on the right where the insulation burned off due to external fire.

 

[G00SE]

@Checkthisout that’s what I think happened as well. That support looks to be a big contributor to the mess

 

[Checkthisout]

@Checkthisout that’s what I think happened as well. That support looks to be a big contributor to the mess

Those supports are notorious for popping out and not being fully installed when you think they are.

That's why I hate those shelves.

 

[hwy17]

I'm opposed to the easily cut theory of the welding wire, this stuff gets so beat up in a welding shop. I'm inclined towards theories that heat that wire and melt the insulation before it shorts to the rack. Also a short to the rack necessarily involves another fault before that, because I doubt OP was running a rack negative grounded system.

 

[Checkthisout]

1) couple questions. If those breaker interrupt positive and negative I think you guys found the cause.

2) If those breakers are only for positive then was the shelving grounded?

 

[G00SE]

I'm opposed to the easily cut theory of the welding wire, this stuff gets so beat up in a welding shop. I'm inclined towards theories that heat that wire and melt the insulation before it shorts to the rack. Also a short to the rack necessarily involves another fault before that, because I doubt OP was running a rack negative grounded system.

I think the leg buckled and then pulled the wire which caused the short