Extenuating circumstances, but EVTV's Jack Rickard found out the hard way that charging a pack with the same cells as the Nissan Leaf that had been dragged down too low without a BMS resulted in a fire that destroyed his workshop.
"This pack was right out of the cargo container and we never even attempted a bottom balance. We were only going to use it for testing chargers and DC-DC converters and the UQM test bench. But as the result of one of our assclowns playing around with the bench while I wasn’t in the shop, it had drained down very slowly overnight to a very low level.
It seemed to charge back up ok. But never quite got to full charge. So I had hooked it up earlier in the afternoon to bring it up some more."
Yes - bad idea to try and charge a pack that is suspicious without any BMS.
"I quickly shut off the charger and cut off the contactors. But it continued to BANG and POP irregularly... Suddenly the pack begins to issue the familiar white smoke – just a bit at first, then more. The pack weighs 450 lbs, and the fork lift is at the other end of the building...."
"By this time the white smoke was coming out pretty good. I don’t know why, but I was curious what the temps were. So I grabbed an infrared gun and shot all the cells. Most were warmish in the 35-40C range but there were two sitting at 95C. Not good."
"Suddenly the pack spewed a spear of sparks and flame about six feet straight out the front – right where I had been a moment before. And then it exploded into a massive fireball shooting flames up to the ceiling with such velocity that they splashed laterally from there."
Lessons to take from this - don't overdischarge, and don't let them get out of balance! If they are out of balance, don't try and charge them without a BMS.
Sorry to hear that happened! But it happened for a reason: they tried to bottom-balance, then over-discharged the pack, then overcharged some of the probably damaged cells. A recipe for disaster with the Li cells.
Yes, a BMS would have prevented both, likely, although the issue is that the Leaf cells have 2s2p configuration in each pack. And there is no way to individually monitor through BMS or otherwise each one of the 2 pouches that are in the 2p part of it: if one of the two "p" cells is bad, the other might compensate enough that the BMS might not notice this and still allow one of the bad cell to over discharge and get damaged, then attempt to charge fully and get the battery up in smoke...
All that said, in my opinion we need to use top-balance instead of bottom-balance for our Li conversions for the Vectrix. And leave a significant safety margin both on the bottom end and at the top end. And never connect anything to only some cells in the pack, which will cause disbalances.
Also, in other non-Leaf conversions, where there are dozens of cells in parallel, the situation is even worse: BMS does not monitor each cell, it monitors a pack of say 20 cells in parallel. One of these cells can easily die and the BMS would not know it (especially a poorly designed one). I think 20 cells in parallel with a few bad ones among them will still cause issues that won't be detected by the BMS and may still result in overcharging...
It's funny , cause this morning i was seeing the episode on evtv , and I thinked to post it here.You have the pole!Regarding the parallels Kocho is referring in a conversion that has 20,30,60 cells in parallel (for example 18650) , it's imperative to tesla-fuse them.
These weren't Leaf cells. Red carefully and you find it's a Renault Fluence pack. Those were the cars Renault sold through Better Place.
Despite Renault and Nissan being in a Partnership (the same CEO running both companies, and lots of cooperation between the companies), the companies use different battery sources. Renault uses LG Chem cells, while Nissan makes their own in a partnership with a Japanese battery maker. That may change with the 2018 Leaf which rumors say will have 300 miles range and possibly use LG Chem cells.
In any case, it means the cells in this pack are probably not packaged the same way as a Leaf pack.
Also - he said this particular pack hadn't been bottom balanced. Therefore the state of charge of the individual cells were in an unknown state, especially after that "assclown" had been dorking around with the pack.
Not that it matters that much to the result.
As the rest of y'all noted, Jack's normal malpractice with battery management caused this blaze. It will be interesting to hear what (if anything) he learns from this, or if he just blames it on the "assclowns" and leaves it at that.
My experience with battery packs shows they rarely have cells that behave perfectly together. Almost certainly some cells will, under load, fall to lower voltage than other cells. And of course for any battery chemistry (even lead acid) there's a voltage below which the cells are damaged, and depending on the chemistry they can even catch fire and explode.
The cell voltages at rest - no load - don't tell you anything about the condition of the pack. That is, except when the cells are fully discharged and at or near their minimum safe charge level. For lithium-ion cells the voltage starts dropping quickly at that point.
With two different vehicles I had a voltage monitoring system displaying per-cell voltages while driving/riding the vehicle. In my Karmann Ghia most of the cell voltages rose and fell in tandem, but a couple of the cells would fall more deeply under load than the others. My duty as the driver was to moderate the throttle to keep that low cell from falling below the minimum safe voltage.
I've seen Leaf owners talk about Nissan giving them a report showing how healthy their pack is, per module, and learning that the pack might be fine except one or two modules.
Basically, while each cell comes out of the same factory and made by the same machines, they're all slightly different.
To know the relative capacity of cells in the pack, you have to watch the voltages as they charge and discharge. The ones with the deepest voltage sag under load probably also rise the quickest while being charged. Those cells have the lowest total capacity in the pack.
If you don't have a BMS, you're flying blind with no protection. You have no knowledge of the condition of any cell in the pack, and you have no protection against whether the cells go into unsafe territory with voltages that are either too high or too low.
- David Herron, The Long Tail Pipe, davidherron.com, 7gen.com, What is Reiki
My Karmann Ghia had (I don't know what the current owner of that car has done) a BMS supplied to me by Lightning Motorcycle. This monitored each cell and - if we had hooked it up - would modify the throttle while being driven to ensure that cell voltages didn't go too low. But I had a monitoring display showing all kinds of details the most important of which was the lowest cell voltage and average cell voltage.
While driving my eyes kept darting to the lowest cell voltage to keep that from falling too low.
While charging it followed a "top balance" strategy. Any cell that rose above a threshold would have a balancing resistor turned on. If the pack voltage rose above another threshold, the charger would be told to stop charging. In that mode the balancing resistors would bring down the voltages of the highest cells, and if the pack voltage fell below yet another threshold the charger would be told to restart charging.
All the thresholds were programmable, and I'd set it to top out at 3.82 volts per cell to lengthen the pack lifetime.
This worked perfectly and for 3+ years of owning that car I could plug it in and walk away and know that it would be correctly charged.
- David Herron, The Long Tail Pipe, davidherron.com, 7gen.com, What is Reiki
The first batchs of renault packs were aesc.The pack burned in EVTV was AESC (AKA NISSAN NEC).
"We managed to get our hands on a Renault Fluence ZE battery pack as part of the Better Place bankruptcy. This pack is basically the same as in a Nissan Leaf, the same cell count, cell manufacturer and basic specifications."
LG come into to game on the Zoe , but twizy , influenza//fluence and kangoo are all stuffed with Nissan Cells.
My bad about the bottom balancing being done, I misread the original posting...
While the layout is different, those cells look pretty much just like the Leaf cells
2015 Tesla S90D
2009 Vectrix VX-1 Leaf Conversion
1983 Lambretta Lynx
1980 Vespa P200
2013 VW Jetta Hybrid (gone....)
This event shows that Lithium cells have to be handle with care.
J. Rickards should have known that charging dead unbalanced cell is very dangerous. If packs was drained under 100V, that means that some cells may have been reversed, so short-circuited.
I use my "leaf" vectrix as Kocho said : charging at 80% and never going down to 15%. This area is very safe and may allow to not use a BMS.
My last control (with good Metrix) showed that there was only 1 mV difference between the more charged and the worsed charged cell. This result is even better than the precision of most BMS !
So I am quite confident but I check regularely the cell voltage.
By the way, concerning Renault EV cars :
- Influenza first generation battery is with AESC modules, exactly the same as Leaf first generation
- Influenza second generation battery, now sold in Korea, is made by LG
- Kangoo is with AESC module, exactly the same as Leaf first generation and second generation then.
- Twizy battery is made with LG modules
- ZOE battery is made with LG modules.
These are good points. Where we are "blind" without a BMS in our Leaf conversions is actively monitoring each cell while riding. I know that near my low SoC, if I measure immediately after riding, I get up to perhaps a 10-th of a Volt difference between the highest and lowest cells. The discharge curve is steep in that region, so differences in actual battery capacity begin to show there. However, even at this point I got plenty of cushion, so even the lowest cells would not be over-discharged. But I usually avoid such low SoC in normal riding. Where I stop and charge is usually at higher SoC, and there I barely detect a few mV disbalance, if any. So no cell is "empty" or below safe SoC.
During charge, I often monitor manually the voltages at each cell. A BMS does that too, under the same conditions I do. So I can tell just as well if not better what the voltage is on each cell (my multimeter is probably more accurate than the BMS). As the SoC rises, the cells that were at lower voltages than the rest when discharged catch-up with the rest. As SoC nears "full", which is 146-147V for my conversion with 18 packs, the cells' voltages all rise together. I can't measure any difference at that point, meaning any disbalances are less than 0.005V, which is insignificant. The charge curve begins to be steep in this region, so if some cells were overcharging, I would see it as their voltage would rise faster than the rest. But so far this has not happened. So, the cells have so far remained top-balanced well since I installed them.
Yes, if a cell fails catastrophically, and I am not measuring to detect it, I won't know without a BMS, so I'll be in trouble. But cells usually so not fail like this, unless there is a disastrously low over discharge or high overcharge or overheating or freezing to cause the damage. None of these events should happen in normal operation - we got voltage protection limits on both charge and discharge, and the temperatures in my use are mild.
Note, that folks with ESD chargers should be mindful of the known bug that causes the charger to occasionally lock-up during charging and not recognize the battery is full, and *can* cause overcharge! The high voltage protection won't kick-in in this case!
Premature loss of capacity due to abnormally fast aging on the other hand, I hope to be able to catch through periodic manual measurements. Since I got the BMS leads easily accessible, it takes less than a minute to manually measure the voltages of all cells in the pack, so I tend to do it quite often.
Would I prefer to have a good BMS? Sure! But they are quite costly and I don't want to spend that much on the Vectrix. And I'd rather not rely on a poor quality cheaper one. I might eventually build a passive monitoring system for per cell voltage that I can turn on occasionally during riding or charging, but keep off the pack's circuit most of the time to avoid it causing disbalances.
I read this with interest but I still run my Vectrix without a BMS with no problems. Having said that, the current software will not allow a dangerously low voltage to occur. TheLaird put a lower cut off limit of 120v on it for me (I believe the publicly available software is higher than this) - prior to this I was going as low as 115v without any problems but I don't recommend it. This winter I am going to remove / clean and test my installation but I still won't be putting on a BMS. I assume the problem here was unbalanced / extremely low voltage cells - once they are balanced and kept charged correctly they are very stable. I'll update my previous post with my updated finding when I get around to it.
Electric traction is the future.
You can't run a Li-Ion battery flat and then recharge it without checking each block of cells. Some will be slightly damaged of if reversed totally damaged and unable to accept a charge. This will cause the good blocks to overcharge.
Jack Rickard has damaged many of his battery packs this way including several batteries in the VW Thing.
Can someone explain what a "bottom balance" is, and how it is done? The only state of charge where "balance" can be certain is the "top" i.e. all cells charged at a constant current until it rises rapidly to the recommended full voltage, then the same constant voltage for all cells until current drops to 0.05C or so. There is nothing like this at the bottom of the SOC, so how is a bottom balance done?
Of course the bottom (empty) voltage curve of a Lithium-based battery chemistry has a similar knee and drop off into oblivion as it has the spike at the top, so you can balance the cells to a point just beyond the knee, so below for example the extremely flat discharge curve of a LiFePO4 cell. Like this the likelyhood of completely discharging to 0 (and potentially destroying) the smallest cell in a bottom-balanced pack is greatly reduced.
The trouble is then how to keep the smallest cell in a bottom balanced pack from completely overcharging when full? I think the Rickard-solution to this is to considerably reduce the charger shutoff voltage to ensure that the smallest cell is not overcharged. In a car battery with 40 cells and more in series that might be a viable solution, but for a scooter with just 20 or 24 cells it suffers noticable performance reduction with every volt of battery voltage less, so I prefer the top balance for my scooter/motorcycles...
2017 Zero S ZF6.5 11kW, erider Thunder 5kW