Detecting LFP cell failure

I've built my own DIY battery pack and it's been functioning beautifully over the past few months. 16 cells generally all within 0.01v. However, the battery is in an off-grid remote location nearly 2 hours away and if something goes wrong I'd like to know ASAP. I do have monitoring with VRM and have access to the BMS via a "Peter board", so I can see everything remotely.

I'm wondering if there are any symptoms that would indicate a cell failure (or pending failure)? I can think of a cell's voltage significantly deviating from the rest of the pack, but are there any other metrics provided by a BMS that could indicate cell/battery failure? Does a venting cell release hydrogen? Could you use something like an MQ2 sensor with an Arduino to detect cell venting? Has anyone done that?

A few things...

Significant deviations between 3.10 and 3.35 are very concerning. Normally, even imbalanced cells will report nearly identical voltages in the operating range at rest, under load or being charged.

A single cell that hits high voltage cut off AND low voltage cut off is the weakest cell, and MAY indicate a problem.

A single cell that consistently lags the others during charge may have excessive self discharge and may indicate premature failure.

Not sure what it vents, but that's an interesting idea. I suspect that any cell that vents might not do so in meaningful concentrations for detection. Most venting seems to include a visual indication, so if you already have visibility of the cells, you may already have means of detection.

I think voltage drift is the most obvious symptom. Loss of capacity will show up as voltage drift in most cases near the top and bottom. Vented gases do not contain much hydrogen. It smells more like ether.

sunshine_eggo said:

A single cell that hits high voltage cut off AND low voltage cut off is the weakest cell, and MAY indicate a problem.

Click to expand...

This is what I'm after, really. What (if any) indicators are there that something is wrong so that the cause can be addressed before the pack explodes and burns my cottage down.

Ampster said:

Loss of capacity will show up as voltage drift in most cases near the top and bottom.

Click to expand...

This makes a lot of sense! I suppose the same sort of thing would happen if you built a battery with different sized cells, e.g. throw a 100ah cell with a bunch of 304ah cells and you're going to see the voltage of the 100ah cell peak far before the 304s, and the opposite when discharging.

I think that it's probably also a good practice to inspect the cells every so often. I think the frequency would depend on your environment, where you'd want to check more often in a mobile environment vs. stationary. Checking the vent caps, looking for any bulges or other physical problems. Also checking the IR of individual cells as well as checking the connections across all of the bus bars and terminals.

About 1/2 of the gasses vented are made up of hydrogen, around 1/20th is oxygen They rest is a soup that you wouldn't want to breath but is not flamable on their own. This combination depends on what the individual vendors do but only varies in percentages and traces. There is a writeup on this topic where someone vented various cells and measured how long it took to fill the volume of a 20ft shipping container to a level where it would burn. The answer was around 20 second in most cases.

So long as there isn't a spark it will all disipate and you just would be left to cleanup.

The pack won't explode, just the hydrogen will burn if it vents and there is a spark to ignite it.

We have kicked around various ideas and in a remote location like that you could install and ardiuno board or other controller linked to a hydrogen sniffer. When it is detected trigger a DC contactor to turn off the pack and also trigger an ignition safe fan to vent things outside. Simple ignition safe is where the motor is outside the room and the shaft comes through the wall to turn an exhaust fan. Or they make small fans similar to a bathroom fan, but with a sealed motor so no sparks escape.

If the records from the house burned down thread are consistent there will be a current spike then a drop when the vent is happening.

Another sure way would be to tape a temp sensor to the side of each cell - 1-wire sensors in series - and hook to a RPi or Arduino to trigger the fan and contactor like above anytime the temperature of any cell radically raises compared to the others. All vents make the cells red hot to the touch.

I forgot about temp sensors. My Pytes batteries have them.