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

CzerwonaKrowa said:

Is this not *DIY* Solar Forum for a reason? No thanks, I'm an Engineer myself (M.Eng in Computer Engineering) and I'd rather not quadruple my costs because of FUD. I am not aggravated at all, I'm just looking for suggestions or improvements or ideas to improve what I intend on building. There are already lots of useful suggestions that I appreciate.


I think I need to clarify some things because I don't think everyone understands this layout and how shorting would manifest in my setup. It is certainly no worse than any other layout and unless I am mistaken it's better/safer than most other layouts in fact.

View attachment 216258

Here is a view with the cement fiber boards in place -- You'll note that each string has essentially a concrete wall/well separating it from each other string. Maybe a metal C clamp could be dropped and wrap around that wall to short between the strings, but even then, as in the above picture, the voltages in the <-> direction are the same between the strings or maybe at most a few millivolts off. Wouldn't be a problem.

View attachment 216259

Here I have added busbars to clarify a bit more how these are wired

View attachment 216260

Finally the "bad short" example is no different than anybody else's setup -- a 1 cell 3.2V short. For the "ok short" example, in most cases nothing would happen, unless maybe 1 string was discharged and out of service and another string fully charged, but still that would be a very unlikely scenario considering all the other protections. There is another scenario where a diagonal short could happen which would potentially end up with the string Class T's blowing and BMS'es disengaging but I think the worst scenario is still the 1 cell short since there's no fuse or BMS to disconnect such a short. But this kind of short risk is the same in all battery systems.
View attachment 216261

Finally, lets say I need to service string #3, I just pull the single panel I need to access the cell I need to swap, put 2 pieces plywood down to shield the strings I'm not working on, and then I can VERY safely work on just the one string. Here is a visual:
View attachment 216264

Also, @DIYrich don't get me started on the volts vs amps "debate". The volts determine how many amps go through you (given your body's resistance). While it's the amps that kill you it's the volts that determine how many amps will flow through you, and in this case 48V DC is considered quite safe to work with. Also, that fact has nothing to do with the layout of my system and applies to ALL battery energy systems.

Click to expand...

How are you handling the compression and expansion properties of the cells given they should be compressed on their wide faces?

CzerwonaKrowa said:

@wpns I am absolutely open to and entertaining a shelf option. I have looked at lots of peoples setups and systems all around this forum and tons of solar youtubers, and I don't like their arrangement of the cells on the shelves. If you want to suggest a specific one that you like I can look at it and maybe voice my concerns about the arrangement that maybe you can then address possibly could start a useful dialogue about that.

Click to expand...

Sorry, the whole thing reads like bad science fiction to me. At that scale (why?) I’d really want a professional COTS solution. Several folks here have pointed out the downsides to the ‘batteries under the floor paradigm’ but it’s your project/fantasy, I’ll watch for build pictures.

If you could slide the batteries out one end to work on them I might be OK with it, but sitting/lying on the floor to work on stuff under the floor??? Nope.

Culchee said:

At the end of the day..this is your idea for a build and have reasons for your approach to the build...what I'd like to ask is why have them underneath the floor...?

Click to expand...

Optimization. Minimize cost, maximize safety and usability.

Culchee said:

If one of your requirements is easy access,then another approach could be to utilise ancient technology..that being the wheel...for example racks that give you access at a workable level...that can be unplugged using an Anderson socket/plug,and wheeled out to work on giving you access to all sides..then wheeled back into place...

Click to expand...

Not only does racking at least double if not triple the cost, but it requires me to handle ~300 lb heavy metal boxes. I also do not like the rackmount batteries due to how difficult it is to replace a bad cell inside one and how the case is made of metal and in case of a cell failure shorting to the case is virtually guaranteed and cascading the cell failure to the other cells is virtually guaranteed (including more shorting to accelerate the disaster). Other things I don't like about racking and rack mount batteries is the way they are wired. I am going to be length matching individual wires from each bank so that resistance to each string is the same so they all charge/discharge the same. Most people's rack mount systems (including Will's) have all the units at different SoC's because the units further from main connection point have more resistance.

Brucey said:

How are you handling the compression and expansion properties of the cells given they should be compressed on their wide faces?

Click to expand...

There is no compression requirement. All that compression does is increase max cycle count, which I'm not too concerned with. The expected cycle count for clamped is ~4000-6000 and unclamped ~2650-4000. Taking the worst case, 2650 cycles -- my house would drain 1 "cycle" from this 274kWh battery store in 5 days. 2650*5/365 = 36 years to fully "use up" the battery through cycling. I think we can all agree that the bigger concern is calendar aging and time-based degradation rather than cycle-based degradation. Clamping is a ton of extra work and a lot of extra expense, the opposite of the goals of this design. I feel like most people clamp cause they see everyone doing it or are told by everyone to do it rather than doing the calculations to see if it's even worth doing for their use case. Finally, clamping also prevents being able to visually see if a cell is puffing up abnormally which is the main way I have known most of my batteries in phones and laptops are starting to fail before anything gets critical.

wpns said:

Sorry, the whole thing reads like bad science fiction to me. At that scale (why?) I’d really want a professional COTS solution. Several folks here have pointed out the downsides to the ‘batteries under the floor paradigm’ but it’s your project/fantasy, I’ll watch for build pictures.

Click to expand...

The scale is what it is due to an analysis of the past two years of power consumption of my house and worst-case figures from worst-case times (winter) to go completely off grid. I don't want to run generators in the winter like most do. A "professional" solution would be prohibitively expensive (~4x the cost) and a huge risk for serviceability and longevity (warranty only applies if the person who installed it exclusively services the equipment, the chance of them still being in business 10-15 years later is highly questionable and then i'm boned, etc.). Currently, this solution has me at below break-even costs compared to just staying on the grid over the long term. Spending 4x the cost makes this no longer make sense to do at all. At that point I skip the ESS completely do solar only and pray the net metering agreements don't go south like they did in Cali but also pray the cost of electricity doesn't rise too much in the future (lol...).

Q-Dog said:

If you could slide the batteries out one end to work on them I might be OK with it, but sitting/lying on the floor to work on stuff under the floor??? Nope.

Click to expand...

When the strings and system is being built initially I would be sitting on the under-floor too since there wouldn't be batteries everywhere yet. I would probably only need to work "under the floor" for the last 6 strings or so. After that it's mostly only cell swapping maintenance to be performed or maybe some bolt tightening / torque spec verification. With this layout swapping a cell is removing 4 bolts, slide cell up and out, slide in new cell down and in, install and torque 4 bolts. Probably a 2 minute operation. I don't see how this is so bad compared to working on a car in the engine bay or in working in an attic installing wiring for example.

Why the need for 274kWh? Wouldn't 60kWh and a backup generator for recharging also work and be cheaper?

CzerwonaKrowa said:

Is this not *DIY* Solar Forum for a reason? No thanks, I'm an Engineer myself (M.Eng in Computer Engineering) and I'd rather not quadruple my costs because of FUD. I am not aggravated at all, I'm just looking for suggestions or improvements or ideas to improve what I intend on building. There are already lots of useful suggestions that I appreciate.


I think I need to clarify some things because I don't think everyone understands this layout and how shorting would manifest in my setup. It is certainly no worse than any other layout and unless I am mistaken it's better/safer than most other layouts in fact.

View attachment 216258

Here is a view with the cement fiber boards in place -- You'll note that each string has essentially a concrete wall/well separating it from each other string. Maybe a metal C clamp could be dropped and wrap around that wall to short between the strings, but even then, as in the above picture, the voltages in the <-> direction are the same between the strings or maybe at most a few millivolts off. Wouldn't be a problem.

View attachment 216259

Here I have added busbars to clarify a bit more how these are wired

View attachment 216260

Finally the "bad short" example is no different than anybody else's setup -- a 1 cell 3.2V short. For the "ok short" example, in most cases nothing would happen, unless maybe 1 string was discharged and out of service and another string fully charged, but still that would be a very unlikely scenario considering all the other protections. There is another scenario where a diagonal short could happen which would potentially end up with the string Class T's blowing and BMS'es disengaging but I think the worst scenario is still the 1 cell short since there's no fuse or BMS to disconnect such a short. But this kind of short risk is the same in all battery systems.
View attachment 216261

Finally, lets say I need to service string #3, I just pull the single panel I need to access the cell I need to swap, put 2 pieces plywood down to shield the strings I'm not working on, and then I can VERY safely work on just the one string. Here is a visual:
View attachment 216264

Also, @DIYrich don't get me started on the volts vs amps "debate". The volts determine how many amps go through you (given your body's resistance). While it's the amps that kill you it's the volts that determine how many amps will flow through you, and in this case 48V DC is considered quite safe to work with. Also, that fact has nothing to do with the layout of my system and applies to ALL battery energy systems.

Click to expand...

Sorry you took my suggestion the wrong way.

You have some serious code considerations. As a mechanical computer engineer you know that utilizing qualified design professionals will go a long way in getting a project done correctly and to code. No FUD here.

Best on your project.
Thanks

Mike

Browse said:

Why the need for 274kWh? Wouldn't 60kWh and a backup generator for recharging also work and be cheaper?

Click to expand...

4-6 days of energy for those long strings of days of overcast skies in the winter for a home that consumes about 50kWh a day. I'd rather stay on grid than use a generator. Home is fully electric, no gas. As an additional note, the system was sized based on both calculations and a much smaller scale 1-year-long experiment with 1380 watts of paneling and a 14.4 kWh battery. The 274kWh size/scale is ultimately what meets my needs and it is what it is.

N7QX said:

Sorry you took my suggestion the wrong way.

You have some serious code considerations. As a mechanical computer engineer you know that utilizing qualified design professionals will go a long way in getting a project done correctly and to code. No FUD here.

Best on your project.
Thanks

Mike

Click to expand...

I don't think there are any allowable "code considerations" for a cell-level DIY LifePO4 battery bank. The code simply does not allow it or have provisions for allowing anything like it. Any professional I could get to design this project would have to use UL-listed certified hardware such as certain rack mount batteries that I simply do not want to use. There are plenty of people on these forums building cell-level packs and while I appreciate what you're trying to say, your statement should apply to all cell-level builds. Are you saying you are against all cell-level builds? If you know of a qualified design professional that somehow has the ability to design and build on a cell level, please give me their contact information and I will legitimately pursue contacting them and exploring that avenue, but I sincerely doubt it exists.

Also, I am trying to conform to code as much as I can -- that's one of the reasons why the building will have 2 exits with panic bars, that's code for a electrical utility outbuilding. I have the code book and I've paid for and been going through Mike Holt's code training material for both general code and solar-specific.

Finally, I posted this thread on "DIY LiFePO4 Battery Banks" for a reason. If I wanted to use rack mount premade batteries I'd be talking in the "Beginner Friendly Plug-n-Play Lithium Batteries" section.

CzerwonaKrowa said:

I don't think there are any allowable "code considerations" for a cell-level DIY LifePO4 battery bank.

Click to expand...

I'm going to assume you're in Poland based on the nick. Please disregard this if not.

If you're planning on putting this structure below the ground (as in it is not raised) the code is that battery banks have to be installed raised 30cm off the floor. Don't ask me why 30 and not 32. I guess 29cm of flooding is pretty likely. 31 not anymore in the entire country - flood plains and mountains included.

Anyway, that's why I'm raising my cells 35 cm above the floor.

Why would you care about code if you keep it off grid on your own property? Imagine one day you want to sell it, or insure it etc. Or it burns down and someone dies (imagine a thief breaking in and making a mess, dying in process). Then you get prosecuted for negligent homicide(OK it is a bit far fetched).

It is your choice of course.

CzerwonaKrowa said:

The code simply does not allow it or have provisions for allowing anything like it. Any professional I could get to design this project would have to use UL-listed certified hardware such as certain rack mount batteries that I simply do not want to use. There are plenty of people on these forums building cell-level packs and while I appreciate what you're trying to say, your statement should apply to all cell-level builds. Are you saying you are against all cell-level builds? If you know of a qualified design professional that somehow has the ability to design and build on a cell level, please give me their contact information and I will legitimately pursue contacting them and exploring that avenue, but I sincerely doubt it exists.

Click to expand...

You're right I think. All the "qualified professionals" I ever met are either trying to make a buck selling ready made solutions or they work for companies that provide these ready made solutions. As a small time investor looking to spend <$50k you're not getting a cell level design with a custom building etc.

CzerwonaKrowa said:

Also, I am trying to conform to code as much as I can -- that's one of the reasons why the building will have 2 exits with panic bars, that's code for a electrical utility outbuilding. I have the code book and I've paid for and been going through Mike Holt's code training material for both general code and solar-specific.

Click to expand...

That's always good to hear

CzerwonaKrowa said:

Finally, I posted this thread on "DIY LiFePO4 Battery Banks" for a reason. If I wanted to use rack mount premade batteries I'd be talking in the "Beginner Friendly Plug-n-Play Lithium Batteries" section.

Click to expand...

Personally I don't like the floor idea for reasons listed. IMO it makes sense only if you need the wall space for something else. If not, why not do this:

Just scale it up. There will be a sheet steel cover spaced appropriately to redirect vent gases. Also cells are all insulated from any metal with the usual fr4 material.

If you don't like metal, why not use the non conductive building materials, just putting cells into the walls instead of the floor? It will be much easier to work on them.

Also, regarding compression it is a matter of personal preference, but I have to say my opinion is directly opposite. Has there ever been a battery fire of a properly compressed cells shown on this forum? I certainly haven't seen one. All I saw were cells stuck together with tape (little compression - the rv thread), cells spaced apart with no compression(house burn down thread) etc. I believe when a fire is caused internally and not due to a short to a case it is almost always for one of 3 reasons:
- manufacturing fault (separator damaged at the factory etc) then as the cell expands and contracts that fault get worked and eventually causes and internal short and runaway. If the cell was compressed sufficiently to prevent relative movement it could last longer
- user error (charging when too cold, overvoltage etc) - not much can help there
- physical damage - not much help with this too.

The idea with compression is to prevent relative movement inside the cell. This does prolong the life of the cell. IMO, anything that prolongs the life of the cell also makes the probability of failure lower.

However, it is fine to differ on that point. Many people swear by spacing cells apart.

@Luk88 Thank you for the detailed and considerate post. I'll make the building above-grade. Yes, I am Polish. The "house burned down" thread cells were in fact compressed but regardless I believe cell compression is purely to extend cycle count, not to prevent catastrophic cell failure. I plan on replacing the cells when they have reached the end of their useful life aka < 80% capacity retention, so maybe 2500 cycles or 30 years (realistically, I'd be happy to get 15 years out of them). Is the pictured stack your build?

That stack method may be viable and I will consider it; however, I don't like that I'd have to disassemble the whole pack to say, change the cell at the bottom. The wall space was originally going to be used to house all of the inverters and electrical panels and might still be done so to leverage the waste heat from that equipment to provide adequate heating for the batteries in the winter. That "house burned down" thread is making me paranoid and so I'm leaning towards having all that equipment in another structure adjacent to this one to save that equipment in case a fire happens in these batteries. Putting all of this in an outbuilding instead of my home was assumed from the beginning and already better than what most people are doing (my condolences to the "house burned down" thread person..)

<edit> In fact, EVE provides cycle life degradation graphs for non-compressed (free standing) and compressed ("fixture"). Important to note though is that they seem to only have actually tested ~750 cycles and the rest is purely a prediction curve. Interestingly, initial degradation is much higher for the compressed/"Fixture" option. Note that these are 1C discharge 1C charge so 280A tests. Due to much lower rates I will be putting these through in my set-up, I can expect better cycle count performance than even this I believe.

I probably shouldn't have used the term rack...I was thinking more of a plywood box that is on wheels...I agree that the various metal box systems that are available would be expensive..have you seen "Ray builds stuff" YouTube channel..he has a very practical and affordable approach to housing batteries..his is also a big system...although he is in Texas I believe..I would recommend watching his battery builds..he is very thorough in every aspect of a diy build.

Regarding the cables from each battery string...it is possible even with a rack or box system to ensure that all cables are of the same length feeding into the main pos and neg terminals..

CzerwonaKrowa said:

@Luk88 Thank you for the detailed and considerate post. I'll make the building above-grade. Yes, I am Polish. The "house burned down" thread cells were in fact compressed but regardless I believe cell compression is purely to extend cycle count, not to prevent catastrophic cell failure. I plan on replacing the cells when they have reached the end of their useful life aka < 80% capacity retention, so maybe 2500 cycles or 30 years (realistically, I'd be happy to get 15 years out of them). Is the pictured stack your build?

That stack method may be viable and I will consider it; however, I don't like that I'd have to disassemble the whole pack to say, change the cell at the bottom. The wall space was originally going to be used to house all of the inverters and electrical panels and might still be done so to leverage the waste heat from that equipment to provide adequate heating for the batteries in the winter. That "house burned down" thread is making me paranoid and so I'm leaning towards having all that equipment in another structure adjacent to this one to save that equipment in case a fire happens in these batteries. Putting all of this in an outbuilding instead of my home was assumed from the beginning and already better than what most people are doing (my condolences to the "house burned down" thread person..)

<edit> In fact, EVE provides cycle life degradation graphs for non-compressed (free standing) and compressed ("fixture"). Important to note though is that they seem to only have actually tested ~750 cycles and the rest is purely a prediction curve. Interestingly, initial degradation is much higher for the compressed/"Fixture" option. Note that these are 1C discharge 1C charge so 280A tests. Due to much lower rates I will be putting these through in my set-up, I can expect better cycle count performance than even this I believe.
View attachment 216377
View attachment 216376

Click to expand...

I believe it was off grid garage that found the following, basically 3000 cycles for uncompressed and more than three times that many with between 4 and 18 psi compression.

CzerwonaKrowa said:

@Luk88 Thank you for the detailed and considerate post. I'll make the building above-grade. Yes, I am Polish. The "house burned down" thread cells were in fact compressed

Click to expand...

I must have missed that. I'm not seeing any springs and were there any cell isolators? I may be mistaken on that too about that thread, but I vaguely remember the author saying "no fr4 isolators were not an accepted practice" when he was building his system. Still, it is only one data point with lots of unknowns.

CzerwonaKrowa said:

but regardless I believe cell compression is purely to extend cycle count, not to prevent catastrophic cell failure. I plan on replacing the cells when they have reached the end of their useful life aka < 80% capacity retention, so maybe 2500 cycles or 30 years (realistically, I'd be happy to get 15 years out of them).

Click to expand...

Fair enough. I have to admit, with the amount of flammable liquid inside I feel much better compressing my cells as per my manufacturer's specs (Higee), but if Eve says it's fine not to compress and they even provide degradation charts for that option I guess it's fine for their cells.

Also, I'll be having a CO2 fire extinguisher there. I know other types are recommended (powder), but I've had an opportunity to fight a carbohydrate fuel fire with a CO2 extinguisher and it worked a treat (of course oxygen depletion in my place is a very real concern).

CzerwonaKrowa said:

Is the pictured stack your build?

Click to expand...

Yes, it's in the process of being built.
Cells have been top balanced(and discharged to 3.15V) few weeks ago. I'm using disc springs on top to dial in compression. I've yet to add reinforcement to the top plate as it bends slightly at full force. Cells are isolated on all sides. There is a corrugated steel cover for the front and the BMS will be on threaded standoffs on a textolite plate in front of the cells. A large fuse will be mounted on the wall above the battery.

Indeed if one cell had to be removed, entire pack would have to be discharged and cells above removed. If I didn't want to compress and I wanted individual cell access I'd probably make individual shelves in this tower. I opted for this tower design to conserve floor space. It is underground in a separate(small) underground space.

CzerwonaKrowa said:

That stack method may be viable and I will consider it; however, I don't like that I'd have to disassemble the whole pack to say, change the cell at the bottom. The wall space was originally going to be used to house all of the inverters and electrical panels and might still be done so to leverage the waste heat from that equipment to provide adequate heating for the batteries in the winter. That "house burned down" thread is making me paranoid and so I'm leaning towards having all that equipment in another structure adjacent to this one to save that equipment in case a fire happens in these batteries. Putting all of this in an outbuilding instead of my home was assumed from the beginning and already better than what most people are doing (my condolences to the "house burned down" thread person..)

Click to expand...

Yes, IMO that is definitely the better way to go. If I couldn't use that option I'd look into automated fire suppression and venting.

CzerwonaKrowa said:

<edit> In fact, EVE provides cycle life degradation graphs for non-compressed (free standing) and compressed ("fixture"). Important to note though is that they seem to only have actually tested ~750 cycles and the rest is purely a prediction curve. Interestingly, initial degradation is much higher for the compressed/"Fixture" option. Note that these are 1C discharge 1C charge so 280A tests. Due to much lower rates I will be putting these through in my set-up, I can expect better cycle count performance than even this I believe.
View attachment 216377
View attachment 216376

Click to expand...

It appears fixtured cells reach 95% capacity after 250 cycles while unfixtured need 500 cycles to deteriorate the same amount. Very interesting. At 1000 cycles "unfixtured" are still winning. At 2000 cycles fixtured start to edge ahead.

Raised flooring systems are a common thing for office buildings, especially in data centres and open concept office setups - look it up.
I have worked with systems that use steel adjustable posts to support concrete 600mm x 600mm fire rated raised floor tiles. These bolted down at each intersection with a special clip that fit into a recess in the corners of each concrete railsed floor tile. There are also mineral wool filled rated floor panels that are a steel pan on their bottoms and factory finished on top for a nice finished appearance. These ones had seals on all edges and lifted out with a glaziers' suction cup handle.
The data centre that had the concrete floor, built the entire subfloor concrete on a 1% slope to each edge, like a sloped parking lot. All the below floor equipment was on galv steel rails on shims to hold them level, and above the concrete level to permit drainage. The adjustable posts were set level on top, to support a finished level floor. The two lowest edges of the concrete sub-floor had continuous trench drains running along the the outside walls at the lowest edge. The idea was any water had a path to the trench drains and the below floor equipment, jct boxes etc were all above the level of the sloping concrete to ensure these would not be submerged.

Sorry, just gonna say it, and echo what others have said, but this is the worst placement for batteries, servicing, and a total waste of space. No way I'm walking on that, while simultaneously having to carry tools for "servicing". This battery is unserviceable, extremely vulnerable, and the rest of the "shed" I consider a waste of space because there is no way i would ever put something even remotely heavy above the "false" floor for obvious reasons.

That being said, I do like the concrete shed idea, and would think that having batteries against whatever wall is "safest" with concern to other stuff around the area, and/or with venting in case there was a catastrophic failure. One would want a path for the heat to go in case it were to "run its course" of a total burn through failure.

The way this layout is done, I imagine catastrophic failure of not just one single side or wall of the shed, but the entire floor and everything above it. While the fire would be contained, it will most likely use everything else above it as fuel.

Once a person (especially an engineer) has put so much time and effort into a plan, they tend to be determined to carry on with it despite the clamor from others. It is a natural human characteristic we all struggle with. Been there.

Now this looks like a proper high kWh install:

Browse said:

Once a person (especially an engineer) has put so much time and effort into a plan, they tend to be determined to carry on with it despite the clamor from others. It is a natural human characteristic we all struggle with. Been there.

Click to expand...

"Perseveration"

Although the video shows a large neat and tidy install, it doesn't address the OP's topic.
To the OP's Topic - Ultimate fire-proof shed
Perhaps the best solution is a separate 'fire proof box' only for the ESS.
Some safe(r) distance away put the inverters and all the AC side.
When I see the cells below an access floor just makes my back ache, but then again, I thought how often do I need to access a battery once it is set up correctly? - not very often it seems from the last year of operation. I don't say I would go this way, but the OP is certainly entitled to experiment and I for one am open to at least review his choice, follow along and see where it take him.

When I see members call up venting and fans to remove combustible gases from an ESS area, I wonder if these systems will just feed a fire with oxygen. In a commercial kitchen for example, triggering a fire suppression system automatically shuts down exhaust and make up air systems for this exact reason. Instead of exhaust systems maybe the OP needs to consider automatic fire suppression and alarm systems.

Browse said:

Now this looks like a proper high kWh install:

Click to expand...

Now THAT is a big bank! Awfully tall though, isnt it?