diy solar

diy solar

Shunt snapped and caused a fire🔥

Not entirely Germain the the OP but we can see how any individual cable in this type of setup could be subject to way higher amperage without a breaker or fuse blowing.

The wires need to be fused or breakered at the battery and then all run into a single combiner box where lead is also fused there.

Thus prevents back feeding into the cables at amperage well beyond what they are fused for.

In server rack batteries, I would make sure and set the individual overcurrent for each bms at whatever the max load on the inverter will be divided by the number of packs.
 
Shunt was right above wood impregnated with flammable glue. If the shunt's plastic melted was dripping fire down on the wood. Put some of that particle wood in a fire see how hot it burns. Reminded me of putting diesel fuel on a fire once it gets heated up burns very high BTU. Probably make great fire starter wood cut it up in small chunks.
 
Shunt was right above wood impregnated with flammable glue. If the shunt's plastic melted was dripping fire down on the wood. Put some of that particle wood in a fire see how hot it burns. Reminded me of putting diesel fuel on a fire once it gets heated up burns very high BTU. Probably make great fire starter wood cut it up in small chunks.

The shunt broke due to high current due to the short. IMO

Probably looked like the air intake heater on a diesel engine for a couple minutes there.
 
Shunt was right above wood impregnated with flammable glue. If the shunt's plastic melted was dripping fire down on the wood. Put some of that particle wood in a fire see how hot it burns. Reminded me of putting diesel fuel on a fire once it gets heated up burns very high BTU. Probably make great fire starter wood cut it up in small chunks.
I have used engine oil and dry sawdust as by-products of some of my work - neither one on it's own burns worth a poop, but mix them together and they make a great fire starter material - try it.
 
This sketch is from earlier in my solar set up - when I ran two inverters, and had five packs.
The packs actually have one more fuse not shown, set between Cell #8 and #9 I put a 200 Amp megafuse. The reasoning being "if all else fails" or since the cells are in heavy plywood cases, if a fire started inside the case, I expect the Megafuse would break the circuit. Maybe, hopefully.
All the rest of the set up is pretty much like this today, just bigger. The main 2P disconnect is now 400A. and I only put one Class T (on positive) and put a Victron 500A shunt in place of the Negative Class T shown in the diagram.
My packs rest in steel angle iron rails in a solid cabinet. Today my packs stack 8 units high with room for one more up top. (I own a forklift, and the ESS is in a steel building with 12' ceiling, on a poured concrete floor so yeah, I just stack them up high to save floor space).
Anyway the combination of breakers and fuses should prevent the type of domino failure we see in the OP's case, or I hope so. If you see a failure point - comment!
 

Attachments

  • Typical Rack Battery Installation.pdf
    399.7 KB · Views: 13
This sketch is from earlier in my solar set up - when I ran two inverters, and had five packs.
The packs actually have one more fuse not shown, set between Cell #8 and #9 I put a 200 Amp megafuse. The reasoning being "if all else fails" or since the cells are in heavy plywood cases, if a fire started inside the case, I expect the Megafuse would break the circuit. Maybe, hopefully.
All the rest of the set up is pretty much like this today, just bigger. The main 2P disconnect is now 400A. and I only put one Class T (on positive) and put a Victron 500A shunt in place of the Negative Class T shown in the diagram.
My packs rest in steel angle iron rails in a solid cabinet. Today my packs stack 8 units high with room for one more up top. (I own a forklift, and the ESS is in a steel building with 12' ceiling, on a poured concrete floor so yeah, I just stack them up high to save floor space).
Anyway the combination of breakers and fuses should prevent the type of domino failure we see in the OP's case, or I hope so. If you see a failure point - comment!

I like the idea of a fuse in the middle of a pack. APC does this on their larger battery banks in the smart UPS.
 
So working in this industry I hate to admit that I've seen a lot of customer mistakes. From what I see in this thread, like others, I would find it nearly impossible that the shunt was the start of this catastrophe. I notice the BMS was a Daly (or equivalent) and among my customers I have had reports of more than 10 of these fail in a unsafe state; both charging and discharging MOSFETs were on.

Clearly, the case of the cells were electrically involved based on the hole in the side of the case. It does not appear proper precautions were made for keeping the cells electrically isolated from one-another, nor for abrasion protection on the larger cables. The control/balance wires look like they were not very cleanly ran. The shunt was at the top of the build and fire travels up very easily and has a very hard time traveling down. It looks to me that it started on the shelves at the bottom and in the process of combusting, the flames melted the insulation off the cables, causing them to short to the metal shelves. Once in contact, excessive current could run through them from the other packs, but the connection may not have been good enough to instantly trip the circuit breakers (which it appears they never did trip).

Even if the shunt was running a significant amount of current, arcing through the air is a much higher resistance path than the pack connected in parallel to it. You don't have ~52v across that gap, you only have the difference in charge level between the disconnected pack and the one next to it. Hard to make a fire-starting arc at 1-3v and a few amps.
 
First off, I'm sorry for your loss; this looks obviously expensive.

A few points:
- A few people brought up the breakers. Although not the root cause here, AC breakers don't work in DC environments. It's not a matter of over-current protection, an AC breaker will trip if overloaded with DC amps, but that's not the problem. It's all about the breaker's ability to extinguish the electrical arc when the contacts are opened under load. DC current can sustain a gap over a magnitude greater distance than AC current can, and DC breakers use permanent magnets to quell the plasma, which is also magnetic. This is also why DC breakers are polarity specific. Someone made a comment they don't "like" using DC breakers... well, this isn't the type of "preference" compatible with popularity metrics.
- What material are the bus bars made from? They look to be aluminum stock from the hardware isle. Were they supported by some means? In the photos they appear to be laying physically on top of cables of opposite polarity. Keep in mind that aluminum under compression, as it would be at the point of a mechanical lug attachment, will compress and over time migrate away from the compression. As the connection loosens the resistance increases, and so these need to have their torque periodically re-checked.
- I have never seen a free-air, base-less, shunt. A shunt like that would have to be designed to bolt into some fixture like one that holds a large fuse, not dangling with the elements under physical tension. Also, all shunts are resistors and by nature will get hot under load. Even under rated load. The photos show them mounted over a combustible surface, no enclosure, no strain relief.

In the next battery array you build I might suggest some considerations:
- Avoid use of permeable materials that are under loading forces. That could be weight load, tension load, compressive load, etc. Wood might make an adequate insulator at 48V DC but that only holds true as long as it remains clean and dry.
- It doesn't always require overcurrent to make hot spots. Think in terms of connections, each current-carrying junction poses the risk of becoming overheated if the tension holding it together diminishes. Aluminum compresses over time, especially with heat cycles, and needs to be retorqued when involved in a junction, so design your system so these junctions are visible and accessible.
- Provide securely fastened routing paths for all current-carrying cabling. Avoid instances where a dangling heavy cable's jacket is exposed to an edge that ends up supporting its weight. The focused pressure on the jacket of a cable passing over and edge will result in compromised cable insulation. Cabling making vertical transitions should be secured to a vertical surface with its weight being carried by multiple clamps/ties as-needed.
- Long runs of DC cabling should have the positive and negative conductors run parallel to each other. This will minimize impedance and voltage loss to inductive currents being created by the current flow in the cable. When parallel, the positive and negative conductors' ability to induce eddy currents in surrounding material is mostly cancelled out.
- Consider putting all-thread into PVC pipe sleaves, or some such material, to that the sharp threads to not pose a chaffing hazard to the cells. Also, use a rigid steal piece of steal, like 12AWG 1.5/8" Unistrut, or similar material, for the endcap span between the two rods. This eliminates the problem of warpage when a permeable material like wood is carrying a mechanical load.
- Enclose all of the DC components and connections to protect them from the environment, and the possibility of any conductive foreign material entering the area.
- Approach design from a "what if..." perspective. "What will this connection affect if gets too hot? What if this wire gets yanked? What if this material is exposed to high humidity or even water? What if something metal nearby topples over near and into this? What if there is a smoke or a fire?" Think in terms of every component having the desire to fail with a radius of destruction around it while doing so.

A regular inspection of all the components under full load is a healthy practice. Keeping everything clean will help here too. I have a FLIR infra-red camera attachment for my phone that works brilliantly for spotting hot spots. Anything getting hot has potential disaster in its future.

I hope it works out recovering from this loss.
 
First off, I'm sorry for your loss; this looks obviously expensive.

A few points:
- A few people brought up the breakers. Although not the root cause here, AC breakers don't work in DC environments. It's not a matter of over-current protection, an AC breaker will trip if overloaded with DC amps, but that's not the problem. It's all about the breaker's ability to extinguish the electrical arc when the contacts are opened under load. DC current can sustain a gap over a magnitude greater distance than AC current can, and DC breakers use permanent magnets to quell the plasma, which is also magnetic. This is also why DC breakers are polarity specific. Someone made a comment they don't "like" using DC breakers... well, this isn't the type of "preference" compatible with popularity metrics.
- What material are the bus bars made from? They look to be aluminum stock from the hardware isle. Were they supported by some means? In the photos they appear to be laying physically on top of cables of opposite polarity. Keep in mind that aluminum under compression, as it would be at the point of a mechanical lug attachment, will compress and over time migrate away from the compression. As the connection loosens the resistance increases, and so these need to have their torque periodically re-checked.
- I have never seen a free-air, base-less, shunt. A shunt like that would have to be designed to bolt into some fixture like one that holds a large fuse, not dangling with the elements under physical tension. Also, all shunts are resistors and by nature will get hot under load. Even under rated load. The photos show them mounted over a combustible surface, no enclosure, no strain relief.

In the next battery array you build I might suggest some considerations:
- Avoid use of permeable materials that are under loading forces. That could be weight load, tension load, compressive load, etc. Wood might make an adequate insulator at 48V DC but that only holds true as long as it remains clean and dry.
- It doesn't always require overcurrent to make hot spots. Think in terms of connections, each current-carrying junction poses the risk of becoming overheated if the tension holding it together diminishes. Aluminum compresses over time, especially with heat cycles, and needs to be retorqued when involved in a junction, so design your system so these junctions are visible and accessible.
- Provide securely fastened routing paths for all current-carrying cabling. Avoid instances where a dangling heavy cable's jacket is exposed to an edge that ends up supporting its weight. The focused pressure on the jacket of a cable passing over and edge will result in compromised cable insulation. Cabling making vertical transitions should be secured to a vertical surface with its weight being carried by multiple clamps/ties as-needed.
- Long runs of DC cabling should have the positive and negative conductors run parallel to each other. This will minimize impedance and voltage loss to inductive currents being created by the current flow in the cable. When parallel, the positive and negative conductors' ability to induce eddy currents in surrounding material is mostly cancelled out.
- Consider putting all-thread into PVC pipe sleaves, or some such material, to that the sharp threads to not pose a chaffing hazard to the cells. Also, use a rigid steal piece of steal, like 12AWG 1.5/8" Unistrut, or similar material, for the endcap span between the two rods. This eliminates the problem of warpage when a permeable material like wood is carrying a mechanical load.
- Enclose all of the DC components and connections to protect them from the environment, and the possibility of any conductive foreign material entering the area.
- Approach design from a "what if..." perspective. "What will this connection affect if gets too hot? What if this wire gets yanked? What if this material is exposed to high humidity or even water? What if something metal nearby topples over near and into this? What if there is a smoke or a fire?" Think in terms of every component having the desire to fail with a radius of destruction around it while doing so.

A regular inspection of all the components under full load is a healthy practice. Keeping everything clean will help here too. I have a FLIR infra-red camera attachment for my phone that works brilliantly for spotting hot spots. Anything getting hot has potential disaster in its future.

I hope it works out recovering from this loss.

Pretty much every point you mention, and some you don't, have already been mentioned.
 
Last edited:
DC current can sustain a gap over a magnitude greater distance than AC current can, and DC breakers use permanent magnets to quell the plasma, which is also magnetic. This is also why DC breakers are polarity specific.
Depends on the specific breakers. For example with midnite DC breakers their smaller ones are polarised but larger ones at 100A and bigger are not.
 
I wonder how long things cooked in there before wife noticed?…I use the Ring C/O Listeners and wifi smoke detectors everywhere in my house and attic. Early detection is crucial.

No one in this forum ever wants to talk about these types of devices, I always get boo’d when I bring up smart fire detection devices..

“Oh what if the internet is out! What if the Chinese hack my house? Give me a break! ….What if your phone pops up a push notification at the first plume of smoke and you actually can get over there and shut it down before it’s too late… that’s what will actually happen… no… Everyone would rather play mad scientist DIY batts with 6 foot long buss bars in plywood boxes.. that’s more fun.. gotcha

2D3E5B55-CE9D-4189-8036-5452FDAC2D63.jpegEDDE1452-05C0-4742-9DB2-5E96601BC234.jpeg08AF84B3-AD2F-4A76-9E4A-43CC32CC2A6B.jpeg
 
I wonder how long things cooked in there before wife noticed?…I use the Ring C/O Listeners and wifi smoke detectors everywhere in my house and attic. Early detection is crucial.

No one in this forum ever wants to talk about these types of devices, I always get boo’d when I bring up smart fire detection devices..
I was doing some research on new smoke alarms for the house and I discovered some weird stuff going on in that Industry. A lot of the detectors on the Market are made by the same company.
Even though they look different and come under well know different brand names, most are just self created competition by one company.
 
The Dewalt shelves that Will also has mentioned are super sturdy and I think would even survive a fire. This isn't great pic of showing all the shelves but you can see a shelf consist of wire rack that sits in a groove with hefty yellow metal braces underneath. In my case I added the extra protection some have mentioned of using rock wool on the sides, top of top shelf, and below bottom shelf. Between the middle shelves is cement board so that if something does happen to cells below a shelf carrying cells itself there is chance it won't affect the shelf above. There is a thick sheet of polycarbonate panel on top of the wire racks to physically protect cells from the wire rack. Polycarbonate also has much higher temp rating then acrylic which I've seen some use. Also note the tubing around the threaded rod to protect cells from any potential physical damage. Threaded rod goes through 1.25"x1.25" angled aluminum on the outside of the plywood since the plywood by itself would not be enough.
1708926410459.jpeg
 
The fact the 4/0 cable lasted long enough to set the shelves on fire twlls me the arc was under 2000A... but using class T on THAT big a bank just seems prudent.
not sure what the instantaneous amp output of 12 optima's wired in parallel are but I have seen the damage caused by the shorting of said bank at full charge. the 4 aught cable (Phoneix Gold fine strand OFC) (date me a bit?) exploded, it did not burn, it exploded about 8 inches of the cable when a careless coworker allowed the cable from the bank to the amps feed interconnect to fall and land on the ground bussbars. for this to be a cable issue the amps had to be low enough to not blow up the cable, yet high enough to crate an arc that instantly started the fire. not sure that it was a short of this sort. though i cannot prove this.. (more like I will not try to empirically prove this.)

this one is a real teaser as it has several avenues that could have caused this. the Op will need to look into it more to be certain.

I will say I agree with either fuses or breakers within 3-6 inches of the battery to prevent short created fires.
 
- Long runs of DC cabling should have the positive and negative conductors run parallel to each other. This will minimize impedance and voltage loss to inductive currents being created by the current flow in the cable. When parallel, the positive and negative conductors' ability to induce eddy currents in surrounding material is mostly cancelled out.
the rest of your dissertation was spot on, but I take issue with this, they should be twisted/paired together to avoid the impedance and voltage loss... at least that is what i was taught many years ago. maybe current theory is different. if so please site references .
 
Back
Top