Battery fire!

Near-disaster struck in the early hours today,when my 36V NiMH battery caught fire whilst on charge. Foolishly I was charging it in the house, but was awoken by the bang of the first cell exploding and was able to get the (very hot and smoking) pack outside.

 

Despite having played with this sort of stuff for years, I learned some very salutary lessons, albeit by nearly burning the house down.

 

1) I blithely assumed that I could charge a NiMH pack at the 0.1C rate without risk, based on lots of experience of charging NiCd packs over the years. Wrong! Unlike NiCd, it seems that NiMH can reach thermal runaway temperatures even at such low charge rates.

 

2) I assumed that two sub-packs of 36V, 3.7Ah, charged in parallel (via a pair of matched diodes) from a constant current source would current share. Wrong! It seems that one of the two sub-packs took more current than the other, as in the charred wreckage it looks like one sub-pack is relatively undamaged, whilst the cells of the other are totally burnt out. To do this safely will mean using two chargers, one for each sub pack, or charging the sub-packs seperately.

 

3) I'd fitted thermistors to monitor pack temperatures, but hadn't yet built the custom variable rate charger. I was using a big constant current power supply, set to charge at 750mA. The total pack capacity was 7.4Ah and the cells were all fully discharged at the start of the charge. I should have fitted thermal fuses or cut outs.

 

Some details:

 

The two sub-packs are each made up from four 12V 3700mAh sub-C cell packs. I don't believe that the batteries were at fault.

 

They had been on charge for about nine hours when the first cell went bang and woke me up. I got up to investigate after the second cell went bang and just had time to get into the study, disconnect the leads and carry the smoking pack outside the back door before it very dramatically caught fire. I carried the pack by the cable and as I got to the path outside the burning cells started dropping out of the pack.

 

The fire was pretty fierce, with the remainder of the cells in the damaged sub-pack exploding quite violently. All that is left of that pack is some charred remnants and a few bits of bent metal.

 

Hopefully my very stupid mistake might prevent someone else doing something similar. Certainly I won't ever charge batteries indoors again, no matter what safety systems are installed. From now on all batteries will be charged on an open bit of concrete, free from all flammable material!

 

Jeremy

Close escape!

 

I see now why batteryspace sell their 36v packs with three separate chargers

Hi Jeremy,

 

Bad luck, and good luck in some respects - it could have been worse.

 

Re your point 1) I had assumed this too, and have even gone to 0.2C for some tiny cells embedded in little gadgets. But on reading further I find the recommendation for NiMH is 0.033 to 0.05 C.

 

Re point 2) In the thread on parallel charging, this is what I had assumed too. Maybe we need to rethink this.

 

So what do you think happened? Your power supply was set to 0.2 C and you didn't have the thermal cut out wired up. For some reason it all went into one pack, so either one cell in that pack went low voltage or one in the other pack went open circuit.

 

Its still only 27 Watts (37 V 0.75 A), which is like holding a soldering iron on it. That's only about 1 W per cell - I'm surprised they can blow up with that. Perhaps the power was all concentrated in one spot. That would require one cell in the burnt pack going high voltage and an open circuit in the other pack.

 

Probably worth a good analysis of the surviving pack to see if anything started there. I can't help feeling that the unanswered question here is why didn't the current go into that one.

 

I think also I am going to revise my charging. Its in a workshop with dial out smoke detector, but the potential damage if a fire got a rapid start is enormous.

 

Nick

As Nick says, that rate was far too high in practice. A third of theory at most is safer long term.

 

They certainly can reach very high temperatures, and as part of my experiments with using up to 128 AA NiMh cells in four parallel groups to make a 37 volt battery, I concluded the test by deliberately running two 32 cell groups into thermal runaway and destruction.

 

One I did with overcharge destruction and fire, the other with high load usage to over-discharge thermal runaway and the same destruction.

 

They had failsafe soldered links, the solder melting and links dropping away to keep some sort of a cap on the result, but the thermal runaway continued within the cells after connection failure and both had to be immersed in water to get the situation back under control.

 

It's why I always insist on using chargers with at least two charge termination methods, and preferably three as in the Shenzen-Powerstream chargers I now use.

.

I suspect that the problem with parallel charging of nimh is that the two packs and connecting diodes need to have identical characteristics within very close tolerance. In particular, the voltage of the packs and their internal resistance would need to be the same for each state of charge in order for them to take 50% of the current each. They would also need to be in the same state of charge all the time.

 

You said that they started fully charged. We know that there is a voltage drop at the end charging. When one pack experiences this drop, it will then take more current and here there is potential for runaway.

By the way, you could add a couple of resistors to each leg to drop a few volts. This would avoid the problems I mentioned but you would have to be able to supply the higher voltage and of course it would consume some more electricity.

It's why I always insist on using chargers with at least two charge termination methods, and preferably three as in the Shenzen-Powerstream chargers I now use.

.

 

Hi flecc,

 

There are a number of ways chargers recognise the full point and terminate, and I agree its good to make use of as many as possible. But AFAIUI what most chargers do at that point is to switch from charging (at 1C or whatever) to trickle charging (at .05 C or whatever) and then maybe time out some hours later.

 

But Jeremy's intention was to be always in the trickle charge mode, so he was only a factor of 2 to 4 over the safe current rather than 20 times over.

 

As I said above, he wasn't putting that much power in, so I'm surprised and worried that the result was so dramatic.

 

Nick

Yes, I understood Jeremy's intention, though my posting concerned more the potential for a disastrous outcome under any circumstance.

 

Having seen the sudden and dramatic nature of thermal runaway, I'm more aware than ever of the need for additional safety margins, and I'm not too surprised at how severe Jeremy's misfortune was. Seen once, never forgotten.

 

I no longer leave any battery charging unattended, as I don't think any can be totally reliable.

.

Edited February 1, 200818 yr by flecc

I've now managed to take a picture of the wreckage.

 

http://farm3.static.flickr.com/2299/2235176774_381d55c5d2.jpg

 

I am sure that more charge went to one pack than the other. The charger was set for about 0.1C, based on the capacity of both sub-packs combined (7.4Ah). The cells were definitely all discharged prior to the start of charging, so nine hours at 0.1C shouldn't have caused a problem. Assuming that most of the charge went into one sub-pack, then this could be a serious over-charge.

 

I'm sure that once the runaway started the cells shorted, most probably as a result of the insulation melting. This is what sustained the heat build up.

 

I've checked the charger and it's still working OK at the set current, so it doesn't seem to be a charger fault, plus the battery was still intact when I disconnected it and took it outside. It blew the end cap off, followed by the ejection of the "good" sub-pack and the subsequent fire and explosions, after I'd carried it outside.

 

The whole incident lasted maybe five minutes, with most of that being me waking up after the first bangs, getting out of bed and dragging the pack outside. The most dramatic part took maybe a minute at most.

 

I'm going to order some more cells and try again, but with the charger modified to include two separate charging circuits.

 

I can't believe just how close I came to burning the house down looking at the final outcome of this incident.

 

Take care folks.

 

Jeremy

Edited February 1, 200818 yr by Jeremy

I'm almost doing the exact same thing as you. I have two sub C, 3800 mah packs in parallel. I just got my second pack yestesday so I haven't had much time to play with them. But I didn't manage to wire it all up and had it on trickle charge last night. In the morning, I charged it up for an hour and that was enough to get me to work.

I had planned on charging the two packs together from now on. But having just heard about your experience, I'm a bit worried I might get the same results. I'm glad I only used the trickle last night and didn't try to fully charge them unattended.

do most people charge in garage/shed rather than in the house?

 

is it really not fully safe to go to bed with say a kalkhoff or ezee battery charging?

I think it would be a good idea to either charge sub-packs separately, or to arrange some form of current balancing system to ensure that charge levels remain within safe limits.

 

The other key thing is to fit some sort of thermal cut out. I'm not sure that I will rely on just thermistor sensing on it's own in future, I will almost certainly fit either thermal fuses or thermal switches, as a back up safety system.

 

My pack was wired with a six way connector, so that each pack and the thermistor string was separately brought out to the charger. I've now checked the "good" sub-pack and it seems to be pretty healthy, with a terminal voltage that would seem to show that it's about fully charged (~40V). As far as I can tell this sub-pack hasn't really got hot, there's no sign of cell swelling or damage to any insulation.

 

The dual schottky diode I used to connect the packs to the charger looks fine too, so I can only conclude that one pack took the bulk of the charge, resulting in the thermal runaway. Once the first cell failed, the rest of those in that sub-pack failed in cascade, most probably as a consequence of the insulation between the cells failing.

 

Apart from the shock and the pile of debris, the only other remnant of this mess is the smell. The fumes from the venting batteries were fairly unpleasant, plus the smoke from the fire blew back into the house through the open back door, as I was busy trying to pour jugs of cold water on the debris.

 

The fire was pretty fierce, although I'm not sure what was actually burning. There were lots of bright sparks, with a very hot yellow/orange flame. The picture doesn't show the debris where it ended up, The cells that were ejected were a couple feet away from the case. I have no doubt that, if left unattended indoors, a very serious fire would have resulted.

 

Jeremy

In order to be legally sold in the EU, ready-made battery packs & chargers whether for a mobile phone, laptop computer, electric bike bike or whatever have to be certified by an independent approval body in order that they may carry a CE mark of compliance. Part of the approval process involves deliberately testing product samples to destruction through overcharging, short circuiting, mechanical shock and high temperatures and in order to gain approval the products must meet pre-defined criteria regarding containment of explosion, self extinguishing of fire, risk to persons and property etc.

 

While the recent news stories about exploding laptop batteries show that the approval process is not foolproof it does give a degree of confidence and there is no reason why charging a CE or UL approved bike battery with the supplied charger should carry any more risk than charging a similarly approved laptop computer, cordless power tool or even a torch or mobile phone. The manufacturers of all of these products rarely or never warn of any need for precautions so obviously don't consider the risk to be serious.

Hi Jeremy,

 

I'm still surprised that 1 W per cell was enough to start it off. But at some stage something must have allowed the stored energy in the batteries to be dumped into the hot spot to really start the conflagration.

 

You say you had Schottkies in, which eliminates one thought I had. I was wondering if one cell in the bad pack had gone and then the good pack forced a reverse current through it. That would have implications for any pack with cells in parallel.

 

Nick

I agree with Ian that consumer battery packs are very safe indeed.

 

Nonetheless, I would never go out and leave a battery on charge in an empty home. The chances of anything going wrong are infinitesimal, but I like to play safe. Likewise I prefer to be awake while charging is in process.

.

I agree with Ian that consumer battery packs are very safe indeed.

 

Nonetheless, I would never go out and leave a battery on charge in an empty home. The chances of anything going wrong are infinitesimal, but I like to play safe. Likewise I prefer to be awake while charging is in process.

.

Agreed. Be safe with all types of battery, but absolutely no need for panic or alarm (Other than the smoke detecting type)

I think that the chance of an approved battery back catching fire is very small. Mine was a home made pack, running in a rather non-standard configuration and without any built-in thermal protection. It's not at all representative of a well-designed and commercially sold pack.

 

The thermal runaway problem seems to be subtle, Nick. As far as I can tell from reading up on it, the cell chemistry is such that the reactions tend towards being exothermic when the cell reaches full charge. There is some deliberate engineering of the anode/cathode area ratio to resist excessive hydrolysis during over charge, but it's quite possible to reach a point where the cell can no longer absorb the excess. The failure sequence seems to be:

 

1) A steep temperature increase as full charge is reached, as the charge reaction is essentially exothermic. The heat output from this reaction can be greater than the instantaneous power input, as the chemical reaction is itself heat dependent.

 

2) Hydrolysis starts as soon as the over-charge condition exceeds the oxygen absorption capability and cell temperature continues to rise. The liberated oxygen and hydrogen increase the internal cell pressure until the vent ruptures.

 

3) Post vent rupture the cell may still continue to exhibit an exothermic chemical reaction, as the stored energy in the cell chemistry can now drive things. If the temperature rise is sufficient to damage insulation (which seems to be the case for my sub-pack) then the additional thermal input from cells shorting together will quickly accelerate the fire.

 

This fits well with what I observed, an initial venting (with fairly loud bangs as the vents let go) followed by a very rapid cascade failure, as the chemical reactions speeded up massively with the increased heat.

 

Curiously, the glass bead thermistors that I had embedded in the packs have survived the fire and still seem to be in spec, although their wires are very oxidised from the combination of the fire and the excess oxygen liberated from the reaction.

 

The alloy case is now extremely soft, as it's been very effectively annealed by the heat! The replacement pack will have a composite case, just to remove another possible cause of short circuits under extreme conditions. I will also look at wrapping the new cells in heat resistant insulation, again to reduce the probability of a cascade failure from cells shorting together.

 

I am reasonably sure that there was no internal sub-pack short or cell failure that initiated this, it seems very much to have been the consequence of an imbalance in charge current between the two sub-packs. I will do a discharge test on the remaining sub-pack tomorrow and see what it's capacity is. This should give me an idea as to the amount of charge it accepted. I can then derive an estimate of the charge that the other sub-pack took which should help to prove or disprove my current theory.

 

She Who Must Be Obeyed is now demanding that I always charge batteries outside...............

 

Jeremy

She Who Must Be Obeyed is now demanding that I always charge batteries outside...............

 

Jeremy

What about charging in a fireplace. That should be pretty safe.

Jeremy

 

I was wondering if this is the first time you tried charging your parallel pack like that, or have you been doing this for awhile and only now encountered this problem?

 

I just charged my parallel pack fully for the first time. I'm at work all day, so I was able to keep an eye on it. My two packs are connected straight together, no diodes, and I'm charging them without diodes as well. I had the thermister connected to the newer of my packs during the charge. Both packs started getting warm about the same time. When the peak was detected, it was my new pack that was warmer of the two. This makes sense since it hasn't had many cycles and therefore should have less capacity and accept less charge. Everything seemed to go well.... for now anyway.

Scare stories cause fall.

 

Hi. I have learned alot from the postings here regarding battery charging. The possibility of fires starting from recharging appliances indirectly caused a nasty accident; before Christmas my aged father awoke in the night & thought he heard a noise downstairs, he had read of fires being caused by Nokia phone batteries & worried that his re-charging phone downstairs might be alight went to investigate. He fell down the stairs & has had a leg in plaster from hip to ankle til yesterday.

Some sort of dedicated re-charging station for the multiplicity of modern gadgets might need to become a standard feature of house design. An uncluttered non-combustible cabinet or counter & an overhead smoke detector rather than a dusty corner surrounded by paper or curtains. I will adopt Flecc's rules about being in the house & not charging overnight.

To try to put the whole thing into perspective, an aquantence of mine is a firefighter and while they don't always know the cause of the fires they are called to he is not aware of any fires caused by batteries. By contrast he has lost count of the number of washine machine and tumble drier incidents, usually unattended at night.

I'll second that, tumble dryers are particularly risky when left unattended.

 

I didn't want to cause any panic with my indications of only charging when being in and awake, and I'm referring to all charging, laptop and other batteries included, not just bike batteries.

 

I just think the sensible position is that if you are in and awake, the very, very slight offchance of something going wrong won't matter, and the same applies to running any electrical appliance.

.

Jeremy,

 

Let us know if you find out any more about the cause. I have added a warning to the thread on Parallel Charging in the Battery FAQs forum. In the long run maybe we should have a summary of the process and the dangers in that forum.

 

How is the house smelling?

 

Nick

As I mentioned before, I don't think that the risk from a properly designed, manufactured and approved battery and charger is really very significant. My set-up is very home-made!

 

A dedicated charging area sounds like a good idea, if I had a fireplace I'd have been tempted to use it. The snag is that the amount of energy in the pack is significant, so you really need some containment to stop exploding cells flying off.

 

I'm currently discharging the "good" sub-pack and it looks like it's fine. Some of the cells at one end have damaged sleeving from the heat, but none are distorted and there is no indication of venting. I won't use this cells in the bike, as I suspect they will have suffered long-term damage, but will no doubt find another use for them.

 

One possible cause is the difference in wire lengths. The dual Schottky was fitted at the charger end of the lead and the sub-pack that failed had wires that were about a foot shorter than the "good" sub-pack. This may have been just enough to ensure that that sub-pack took a little more charge current than the other one, resulting in the imbalance. In some ways, not having the diodes might help, as it would allow self-balancing, perhaps.

 

The failure mechanism once the cells get hot is easy to see. The sleeve on each cell body is thin and as soon as it gets hot it splits and shrinks back. This allows adjacent cells to short, so massively increasing the amount of energy being dissipated. The "good" sub-pack has clear evidence of the way that heat almost completely removes the cell sleeves, as several cells at the "hot" end show severe signs of this, with large areas of cell case exposed. There are slight indications that some of these cell cases shorted against the aluminium case, even though they were wrapped with additional tape (which has melted). Luckily, this sub-pack was ejected from the case by the pressure from the cells in the pack that failed. This is most probably what saved them from also blowing up.

 

The smell has virtually gone, thank goodness, although there is still a big burn mark on the path. I found some other cell debris this morning, bits of perforated metal etc, about 6ft from the main fire. It looks like they were thrown there from one or two of the most severe explosions.

 

I'm considering my options for a replacement pack now. I could go for the same set up again, or I may opt to try something else. Whatever I do the pack will have a lot of built in protection systems.

 

Jeremy

 

In some ways, not having the diodes might help, as it would allow self-balancing, perhaps.

 

 

In my multi pack experiments with AA cells, I had switching for the pack charging so the packs could be charged individually or in groups. I found that when charging two or four packs, the charge self balancing was much better without Schottky diodes, and in the final range extending battery version using 2 x 32 cells in parallel, there's no Schottky diodes and I charge in parallel all the time.

.