As with all matters of safety, to you I am just a random person on the internet writing things, and the following is purely educational content based on my own musings, and is not guaranteed to be factually correct or good safety advice.
If you follow battery news (and I don't blame you if you don't), you'll have heard a bit of a controversy surrounding Lithium Iron Phosphate batteries recently. Also affectionately known as LiFePO4, this is a type of lithium battery, which is currently the most common chemistry for home energy storage, and is quickly taking over the EV market as well. I personally have used them in my EV mower conversion and have bought a whole lot more for other energy storage usage.
To a lot of people, LiFePO4 was supposed to be "safe", because they don't catch fire. And that's true. You can abuse these batteries quite a lot, and they won't erupt in flames like other lithium chemistries. They also don't contain cobalt. And they're quite cost-effective. There's really no disadvantage, other than a slightly lower energy density, which doesn't matter for home energy storage, and can even be tolerated in EV applications. The DIY solar community is therefore in love with them, and the amount that they have been used both by DIYers and in commercial systems does give them a good safety track record.
However, more recently there have been a few headlines about their ability to generate hydrogen gas upon failure and cause explosions. It's not the battery itself that explodes, but the gas they have vented, combined with oxygen from the air. In one example, an entire building was reduced to rubble. Holy s***!
But it's interesting to me, that to date there are only four (4) documented examples of explosions actually occurring, despite the fact that there are literally millions of installations of these batteries all over the world, and many consumer-grade systems do not have any protection mechanism against hydrogen buildup (e.g. sensors or engineered ventilation).
Without getting into the argument of whether we should even worry about this, given that it's on the order of a 1-in-a-million event (suffice it to say I'm not going to lose sleep over the idea of my lawnmower exploding) - why is it actually so rare, and what does that tell us about how to make sure you are not one of those unlucky people?
Firstly, if you have a failure, and hydrogen is generated, that doesn't automatically mean an explosion. It seems you have to hit a very specific set of conditions for one to actually happen, which don't occur easily in most systems. You need a hydrogen concentration of between around 4% and 50%, sufficient oxygen, and an ignition source for an explosion to occur. Ignition sources aren't hard to come by especially given that hydrogen is exceptionally easy to ignite, so avoiding an explosive mixture in the first place is the only reasonable way to make sure things are safe.
But a cell venting inside a very small enclosure may not actually create explosive conditions easily. The vented gas creates high hydrogen concentrations, but quickly displaces too much oxygen. On the other hand, a small amount of venting inside a large room may never reach a sufficient hydrogen concentration to be explosive, especially in most non-airtight buildings where the hydrogen will slowly escape anyway.
Surely, then, there are simple mitigations that would greatly reduce the risk of ending up with explosive concentrations if a cell failure were to occur. Besides operating everything outdoors, which isn't always practical, simply making sure you ventilate your system to a large enough room might be enough. In which case, the primary hazard probably isn't hydrogen explosions - it's breathing in the other toxic gasses that the batteries also release. Hydrogen also is lighter than air and will rise to the top of a room - so if you make sure your ceiling is at least slightly porous, it should find a way to escape instead of collecting in the room.
Secondly, you need a cell failure - a thermal runaway - to actually generate hydrogen from these things. LiFePO4 cells don't produce any gas during normal operation. And it's harder to push them to thermal runaway, compared to other lithium chemistries. With all that said, overcharging them, overheating them, or mechanically damaging them could potentially cause gas release, as could an underlying defect in cell construction that results in abrupt failure at some point in the cell's life. Usually a properly designed system with a battery management system (BMS) will be very good at preventing all of these things, and most cells from reputable sources these days are well made, so offgassing is exceptionally rare anyway.
So I think the reason this is extremely rare, despite clearly being physically possible, is the rarity of LiFePO4 cells actually entering thermal runaway, combined with the fact that specific concentrations of hydrogen and oxygen are required for explosions to occur. But I do think it is a risk that needs to be considered, mostly when you put a large system (multiple kWh) indoors.
How do I design a system that won't explode then? Can I guarantee my system will never explode?
Practically, there are few guidelines about how much hydrogen a failing cell could produce. Estimates range from 25 litres per kWh to 250. However, researchers happen to have actually scientifically tested the specific 105Ah cells I use (I mean, they don't name the manufacturer, but the specifications, dimensions and weight given in the article exactly match the EVE LF105 cell). They found that they generate exactly 43.8L of hydrogen when pushed to thermal runaway, equivalent to about 130L per kWh.
It's very tempting, then, to use the total installed capacity in kWh, calculate the volume of the room the cells are in, and determine whether a partial or complete failure would create explosive conditions. But there's a problem with this method.
For example, a 4m by 3m by 2m room, coincidentally the same size as my bedroom, is 24 cubic meters, or 24,000 litres of air if the room was empty. It isn't empty, especially when my over-inflated presence is inside it - so let's conservatively estimate that there are 18,000 litres of air left (I'll assume me, a few pieces of furniture and some equipment are somehow 6000L in volume). It's tempting to say that I need over 4% of that, or 720 litres, of hydrogen to make it explosive. That would mean I can install up to 5.5kWh of my LiFePO4 cells in that room, and not expect it to become explosive, even if all the cells failed completely (the other fumes would still likely be lethal).
The problem is that I assumed the gas would mix completely evenly with all of the air in the room. That's actually not very realistic, as I don't have a massive fan constantly circulating the air, so the hydrogen will probably go up to the ceiling and only diffuse slightly. As soon as any part of the room is at an explosive concentration, there is a risk - so this isn't a valid method of establishing safety. The scientific article I linked above shows the type of simulation that would have to be undertaken to conclusively determine how much hydrogen you could release into a given room before an explosion risk would be present. Basically, it's not just how much gas, but where it goes and where it collects. It's quite complicated, to me at least.
I think it's simply unrealistic to expect that every single LiFePO4 battery installation, commercial or DIY, would be accompanied by such a detailed gas-diffusion simulation. For standardised designs used in grid-scale storage, sure - but for residential and light commercial installs, which are all bespoke due to the different buildings and places the systems are installed, the complexity of that simulation seems like a roadblock. And as soon as some environmental factor changes - different ventilation, more objects in the room, etc. - the results become invalid.
I am of course using batteries made of multiple cells, and they may not all fail together. In my case, the LF105 cells are large prismatic cells, and I might have 8, 16 or 24 cells in a system - but not hundreds. In one of the documented explosion cases, around 2500 small cylindrical cells were used across six battery units, each with a 16s/26p (16 series, 26 parallel) configuration, and it looks like a lot of them must have failed quite badly, which when combined with a very well sealed room led to an explosive atmosphere. One assumption that is VERY tempting to make is that only one cell will ever fail at a given time. But as that case shows, this definitely has its limits. If the entire pack is overcharged, they'll all fail together. If the thermal runaway from one cell propagates to the surrounding cells, multiple cells may vent. And importantly, in the case of a pack like the above, which has 26 cells in parallel, one problematic cell will potentially short out another 25 cells, depending on whether they have individual fusing - which they often don't - hence dissipating a lot more power than it could on its own, creating a lot of heat, and potentially creating a runaway chain reaction with the other 25 cells in its group, which then may become bad enough to spread to other cell groups. The BMS will typically have no way of preventing this, as the parallel groups of cells are hardwired together, and it may not even sense a problem until things are very serious, unless there are temperature sensors on every cell (but who's going to pay for 2500 temperature sensors?).
This is why I think having a relatively low number of larger cells and a series-only configuration is safer. If the cells are high quality and manufacturing defects are exceedingly rare, which is typically true with modern cells from reputable sources, multiple defects occurring independently but also simultaneously is not very plausible. And a properly designed BMS with failsafe mechanisms, such as a mechanical normally-open contactor that can disconnect the battery even if the BMS loses power, combined with a charger that is inherently voltage-limited to a safe overall pack voltage, makes whole-string overcharging very unlikely indeed. The cells, especially larger ones, can also be separated with insulating materials to reduce heat transfer and limit thermal runaway propagation - and LiFePO4 is less likely to do this than other chemistries anyway, so the requirements for this are less than for other chemistries. And finally, a single cell failing in a series string won't short out any of the other cells. In my view, the risk of a safely designed system like this experiencing even a single cell failure is very low - but even that (e.g. due to a cell defect or balancing failure) is still much higher than the risk of multiple cells failing together.
This is where you really shouldn't use any of what I'm saying as design advice - I assume no liability if you use this information and your system explodes.
The gas vented by each LF105 cell would have to mix completely with about 1 cubic metre of air (1000L) to stay below the explosive concentration of just over 4%. The big question, and perhaps the crux of this matter, is how hydrogen mixes with air in a confined space, and thus how much hydrogen can be released into the room without making it explosive. I'm not an expert on this, but the research paper I linked above simulated a 1.42MWh (1,420kWh) system in a 31 cubic metre room which was approximately 50% air (so very conveniently about the same amount of air as my bedroom), and found that one cell failing was well below the explosive limit throughout the room, 24 cells failing eventually resulted in explosive concentrations in some parts of the room, and 48 cells failing made things a lot more dangerous.
Based on these results - and again, this is not advice - it seems that perhaps 2-3 cells could fail within roughly that size of room, and explosive concentrations would not be reached. This is also assuming no hydrogen leaves the room - but hydrogen is great at escaping confined spaces over time. If more cells failed, but they did so at different times, the hydrogen buildup wouldn't necessarily exceed explosive limits. That is again saying nothing of the other toxic gasses and the potential danger to human health, but that's another issue.
A lot of the mitigation methods used in grid-scale energy storage units aren't really feasible in most domestic installations. Installing systems outdoors is great in theory, but in practice makes things complicated due to moisture and temperature variations - ventilating the battery enclosure without allowing moisture and contaminants (read: spiders) to intrude is complicated. Another method is to have systems that deliberately create arcs to combust flammable gasses before they reach explosive concentrations - but you're not likely to see that method in domestic installations.
Hydrogen sensors appear to be readily available, and indeed ones that are low cost (under $20 NZD) and which can be readily integrated into a circuit design. This means it should be feasible to design some kind of gas detector, which could be connected into the BMS or used as a standalone warning system. This is more than an ambulance at the bottom of a cliff, because these sensors can detect low concentrations such as 0.1%, which are still well above normal levels in air, but well below the flammable limit, and thus provide some kind of early warning. I would imagine that one of these sensors installed on the ceiling of a room containing LiFePO4 cells, and connected to monitoring, audible alarms and a way to shut down the battery system would be a good safety measure.
Am I going to freak out and get rid of all my LiFePO4 batteries? Absolutely not. Am I going to put them all outdoors? Also no. But I'm aware now, that if I put a 5kWh battery system based on LF105 cells in a small room, and more than 2-3 of the cells fail and vent, I will probably create an explosion risk. Just like I'm aware that a single cell defect or overcharge in any normal household item powered by one of the more flammable lithium chemistries could burn the building down - so let's just put this in perspective! It's definitely something that should be considered when designing or installing LiFePO4-based battery systems indoors - be that DIY, commercial portable power stations, or hardwired systems installed by electricians. How much capacity is installed? What are the potential failure modes? How likely are they? How much hydrogen would be generated by them? Where would that hydrogen go - and would it create an explosion risk somewhere? But I don't perceive any of this to be a large risk compared to other risks that exist from technology, such as those posed by low cost, unregulated lithium batteries of other chemistries in e-scooters and powerbanks, which have the potential to cause fires. And the data, which suggests that this is on the order of a 1-in-a-million event, should put people's fears about it in perspective.
But battery safety does require a bit more nuance than the "LiFePO4 is safe" message that has generally permeated the hobbyist community in recent years, even if in most cases that is very true.
None yet!
Display Name
Email (optional, not displayed)
Website (optional, displayed, include "https://")
Comment (up to 1000 characters)