diy solar

diy solar

Design for "ultimate" "fire-proof" 274kWh+ battery shed

Having a BMS fail to isolate is always a concern for a single string or 16 strings in parallel, the risk of damage for 16 is just much higher.
If a single cell is exposed to over-voltage, it could start a cascade of failures - to all the cells in a pack, and all the packs in an ESS. The simple approach would be to monitor every cell, looking for trouble (excess voltage or temperature) and shutting down a pack to provide the operator time to investigate, and avoid letting the problem expand to other cells or packs.
My thought is there should be no reason to shut down the entire ESS of many packs, due to one potential problem BMS or cell, especially since we will may rely on the ESS to power the monitoring system itself.

An overall system to monitor all the cells in all the packs, and provide over all control makes sense, but may be more complex to implement. To keep wiring simple, the overall monitor would rely on comms with each pack, but those comms are supplied by the BMS's and a failure of a BMS may mean loss of comms - ie no trigger to cut off that high voltage to a single cell - ?
And we don't want a lot of wires external to a battery pack, since this will make service and removal difficult. So how would the external monitor shut down the one pack - it would need a suitable rely, and the relay needs wiring to control it. Or a signal at the very least.

I feel more confident with a secondary system built into each pack, independant of the BMS to trigger cutting the circuit if any cell is out of bounds (cell voltage or temperature).

For excess current protection, I like fuses for each pack, no electronics required, I use a class T between pos bus and each pack, and a 2P breaker to be able to control power to the battery terminal plus a mega fuse inside each pack between cell #8 and #9. (the mega was suggested by a forum member to me and I accepted this was a simple low cost safety device to put into each pack). Over-all I have two fuses and two breakers connected to each and every pack, and a pair of main disconnect breakers and 400A class T's for the ESS as a whole.
 
If a single cell is exposed to over-voltage, it could start a cascade of failures - to all the cells in a pack, and all the packs in an ESS. The simple approach would be to monitor every cell, looking for trouble (excess voltage or temperature) and shutting down a pack to provide the operator time to investigate, and avoid letting the problem expand to other cells or packs.
My thought is there should be no reason to shut down the entire ESS of many packs, due to one potential problem BMS or cell, especially since we will may rely on the ESS to power the monitoring system itself.

An overall system to monitor all the cells in all the packs, and provide over all control makes sense, but may be more complex to implement. To keep wiring simple, the overall monitor would rely on comms with each pack, but those comms are supplied by the BMS's and a failure of a BMS may mean loss of comms - ie no trigger to cut off that high voltage to a single cell - ?
And we don't want a lot of wires external to a battery pack, since this will make service and removal difficult. So how would the external monitor shut down the one pack - it would need a suitable rely, and the relay needs wiring to control it. Or a signal at the very least.

I feel more confident with a secondary system built into each pack, independant of the BMS to trigger cutting the circuit if any cell is out of bounds (cell voltage or temperature).

For excess current protection, I like fuses for each pack, no electronics required, I use a class T between pos bus and each pack, and a 2P breaker to be able to control power to the battery terminal plus a mega fuse inside each pack between cell #8 and #9. (the mega was suggested by a forum member to me and I accepted this was a simple low cost safety device to put into each pack). Over-all I have two fuses and two breakers connected to each and every pack, and a pair of main disconnect breakers and 400A class T's for the ESS as a whole.
Just wondering about the reasoning behind megafuse between cell #8 and #9..does the 2P breaker sit in-between the charge controller and battery pack may I ask...I'm building a diy pack and havent really investigated the fuse side of the build...
 
Just wondering about the reasoning behind megafuse between cell #8 and #9..does the 2P breaker sit in-between the charge controller and battery pack may I ask...I'm building a diy pack and havent really investigated the fuse side of the build...
On my DIY packs, I put a 2P 125A breaker, electrically it's between the pack terminals and the BMS output (on the neg side) and between the POS terminal and the main pack POS on the positive side. This serves two roles: first and most often, it provides a way to de-energize the terminals during handling the pack or service work; second and less often (actually never so far) it serves as a disconnect mechanism if more current flows than the pack is supposed to have - I use 100A BMS's.

The Megafuse is an idea a forum member shared with me, I wish I could remember who and provide proper credit, this fuse is 200A and I put it between cells 8 and 9 where the two rows of cells 'turn the corner' in the DIY pack case. The mega allows an easy place for a final "if all else fails" circuit interruption point that will simultaneously cut the pack into two 24volt halves and cut the flow of current. Typical bus bars supplied with the cells are too short for the #8 to #9 location, so the mega serves that role as well.

I also use T-class fuse between every pack and the POS ESS bus bar plus a 1P DC 125 breaker attached to the server rack case beside each pack - again the breaker is to allow an easy convenient way to de-energize one pack for service while all others continue to operate. The class T is in case 'something bad happens' and all the other packs suddenly try to feed a faulty pack with all they got, then the pack class T should trip, or the 1P should trip, or the 2P pack breaker should trip, or the BMS should, or the Mega should. The chances all five fail me in a time of need seems highly unlikely.

The breakers are mostly for my convenience.
As an example: if I want to do something with one pack, I switch off the 1P breaker on the POS first, this isolates the one pack from the ESS POS bus. Then I shut off the 2P on the pack itself (which disconnects the terminals only, but allows the BMS to remain connected so I can check/adjust settings, review history etc with the BMS BT.) Then I remove the cable from the NEG terminal first, followed by the POS terminal, and finally slide out the pack to do whatever was needed. Reinstalling is the same list in reverse order.
 
I finally found the fascinating video I was talking about earlier. Took me hours. I have not seen anything else like this done anywhere else and reaffirms my thoughts to ditch wire insulation and stay away from metallic things in general in this build. Main thing that always catches is the insulation. All of the crazy smoke you see hissing and puffing out at insane rates before turning into blaze in most of these tests is all insulation. Main cause of these events in the field/jobsite is accidental shorting of cables to cabinets, other cables, batteries, etc. by installers. The crazy thing is the system in this video is *this* insane and it's just lead acid, not even lithium. Lithium will be even crazier if things are shorted like they were in the video.

I still need to review all posts and post thoughts/comments, sorry about that, work and home have been keeping me very busy these past few days -- looking forward to the weekend lol..
 
Last edited:
My thoughts in all of this (not specific to this post).

1). All things have inherent risk. Determine likelyhood of the event and the severity if the event occurs. Mitigate risk to a reasonable/acceptable level. If you can spend 1.5x the cost to mitigate that risk to a 95% reduction, great. But that last 5% might cost many multitudes north of that amount.

My risk mitigation includes a power shed away from other structures, so if it goes, it goes. I did use rock wood, 3/4 ply and 5/8 Sheetrock, but going must past that, the lemon wasn’t really worth the squeeze. This mitigation strategy probably wouldn’t work in an urban environment.

2. KISS. These are already complex systems. Try not to make it more complex.

(A digress here):

Which is why I’m very much against things like rapid shut down in my environment. The chances of a firefighter walking the roof of my shop in the event of an fire are about 0.0000000000000000000000000000000000001%. Rural volunteer fire department, wooden trusses on 4 ft centers, ain’t happening.

3. “Perfection is the enemy of progress.”

4. Two is one, one is none.
 
Last edited:
On my DIY packs, I put a 2P 125A breaker, electrically it's between the pack terminals and the BMS output (on the neg side) and between the POS terminal and the main pack POS on the positive side. This serves two roles: first and most often, it provides a way to de-energize the terminals during handling the pack or service work; second and less often (actually never so far) it serves as a disconnect mechanism if more current flows than the pack is supposed to have - I use 100A BMS's.

The Megafuse is an idea a forum member shared with me, I wish I could remember who and provide proper credit, this fuse is 200A and I put it between cells 8 and 9 where the two rows of cells 'turn the corner' in the DIY pack case. The mega allows an easy place for a final "if all else fails" circuit interruption point that will simultaneously cut the pack into two 24volt halves and cut the flow of current. Typical bus bars supplied with the cells are too short for the #8 to #9 location, so the mega serves that role as well.

I also use T-class fuse between every pack and the POS ESS bus bar plus a 1P DC 125 breaker attached to the server rack case beside each pack - again the breaker is to allow an easy convenient way to de-energize one pack for service while all others continue to operate. The class T is in case 'something bad happens' and all the other packs suddenly try to feed a faulty pack with all they got, then the pack class T should trip, or the 1P should trip, or the 2P pack breaker should trip, or the BMS should, or the Mega should. The chances all five fail me in a time of need seems highly unlikely.

The breakers are mostly for my convenience.
As an example: if I want to do something with one pack, I switch off the 1P breaker on the POS first, this isolates the one pack from the ESS POS bus. Then I shut off the 2P on the pack itself (which disconnects the terminals only, but allows the BMS to remain connected so I can check/adjust settings, review history etc with the BMS BT.) Then I remove the cable from the NEG terminal first, followed by the POS terminal, and finally slide out the pack to do whatever was needed. Reinstalling is the same list in reverse order.
Thanks a million....👍
 
I have been lurking on these forums (especially in the up-in-smoke and LifePO4 DIY forums) and been learning a lot and thinking about a design for an unconventional battery shed that will probably stir up a lot of controversy but I have specific design requirements that I will outline and would like to brainstorm as a community to iterate on the design to make it safe, robust, and easy to maintain. Once I am satisfied with what we came up with, I will build it and continue to report on the build.

I will outline the build progressively with more details -- First, a general image showing the overall "vibe":View attachment 216211
The overall idea is to build a structure almost entirely out of concrete and have the batteries be underneath a "false floor' with removable panels for servicing. Each line of batteries is a 17s arrangement of 280Ah LifePO4 cells, with the Class T fuse at the positive end (not pictured) and the BMS at the negative end (pictured). The underneath of the batteries will need some kind of 1" or so subfloor on top of the concrete to redirect moisture so that in case a leak in the structure happens the batteries won't be sitting dipped in water. The false-floor "tiles" themselves will need to be fireproof as well (tile on some kind of cement fiberboard maybe?) and the vertical 2x4's that support them will need to be changed to something fireproof as well, I just don't know what yet. For the wall-side supports I was planning on using steel L channel though I worry about the potential for that to come loose and short out that closet row of batteries. Maybe a little lip detail molded directly into the concrete pour of the structure's wall?

The balance cables for the cells will be made of various gauges of magnet wire and will run underneath the cells in some kind of channel. The image shows them running beside instead of under since that is a design decision I changed/updated/iterated on but haven't updated in the 3D model yet. This was updated to accommodate the addition of the cement fiberboard between each row of cells. The reasoning for some of these choices is that I don't like normal thin insulated wire since the insulation is just flammable fuel in case something unexpected happens and magnet wire still has insulation on it but so little that there isn't much fuel for fire. The different gauges are because in this arrangement the furthest cell is quite far from the BMS and the closest is very close so there will need to be a variety of gauges of the balance wire to keep the BMS happy with the wire resistances to the cells (already tested/experimented with the JK BMS es I have for this and indeed it's necessary). They need to run underneath the cells because I want maintenance to be dead-simple. I don't want to untangle a rats nest of balance wires that run above the cells to pull a single cell out and replace it. The balance wires will come up between sequential cells and attach with an alligator clip to the bar that links cells. A lot of people around here bash on alligator clips for balance wires and that's something I don't understand -- you want the balancing and cell monitoring system to lean towards the failure side. For example, if an alligator clip accidently gets knocked off, the BMS will see 0 volts for that cell and disconnect. It's almost like a trip-wire system in that regard. I prefer to keep disconnect/safety sensory systems on the fail-sooner side than drilling a hole in a busbar for a super-solid connection for a balance wire that should never see more than 2A of current through it.

Each row has 17s LifePO4 (I prefer 17s to 16s in my experimentation because the voltage ranges for my inverters is happier with the slightly higher voltage range from 17s instead of 16s and all of my BMSes are up-to-20s anyway, and if you don't want the 17th cell you can remove it and run 16s instead). Between each row, there is a cement fiberboard (pictured at the very right there for just that 1 row but in the final build all rows will have them) and also a cement fiberboard fragment between each series cell as well (not pictured) with a small slit cut out for the cell-to-cell link to poke through and connect the cells in series to one another. This "interlinking" grid of cement fiberboard is to try to contain individual cell failures (think internal hard short) as much as possible.

The negative end of the pack and the positive end of the pack will be at complete opposite ends of the structure and I am considering using uninsulated wire to link everything to link everything together (past the Class T fuse on the + end and past the BMS on the - end) to a big 1000-2000A copper busbar on each end forming the positive and negative rails, using some kind of fire-proof but non-metallic bracket (not sure what) to mount it to the concrete walls of the structure. I want to avoid insulation since the only voltages in this building are 48V nominal (60V max) so they don't pose much shock hazard and insulation on wires is a huge fire fuel for any electrical faults.

Other thoughts are -- I want the building to have a ~4000-5000CFM explosion-proof all-metal ventilation fan that will turn on if smoke is detected (they cost about $1k) so that if a battery goes and starts venting the explosive vapors are expelled out of the before the air-fuel mixture reaches ignitable levels. The plans are to house all of the inverters and other electrical equipment outside of this structure so that a battery fire that destroys ~$40k of batteries doesn't also destroy another ~$30k of other electrical equipment.

Finally, I still haven't figured out what to use for battery heating for winter time or a general forced air ventilation option to keep air flowing into the under-the-false-floor chamber to keep the temperature consistent or whatnot and there's a lot of other details not figured out yet as well such as concerns about rodent damage and ease-of-inspection. I will answer questions and keep iterating on this for at least a few months before I begin building this. I welcome suggestions and a discussion around the design of this thing.
I like the separate off house building idea. Put some cooling and ventilation as necessary. Don't like having the batteries on bottom where anything can be dropped on them. Everything falls down and personally the older I get the concept of kneeling is less attractive as it once used to be (oh boy that can be taken different ways :fp2)

Have the floor where you can hose it down with some form of drainage outside. Makes it easy to clean up when you spill your "barley pop" on it. You will also be safe from that 1000 year flood that never happens.

Build the batteries on wall and bring cables overhead on racks. OMG, it sounds like a radio/cell site!
 
Back
Top