The LFP BMS Experiment I'd Like to See

[PianoAl]

We've discussed how there's a tradeoff between lessening LFP battery degradation (don't charge to 100% often) versus preventing BMS decalibration (charge to 100% frequently), but I haven't found any information on how bad the BMS calibration might get.

Here's the experiment I'd like to see someone do (perhaps a YouTuber):

1. Keep the SOC between, say 40% and 70% for a thousand(?) miles.
2. Drive until the car runs out of charge.

That may show how "off" the BMS calibration gets. Does it run out of charge when the display is showing 20%? Does the car have lots of charge left even when it is displaying 0%? Does the displayed SOC suddenly change as it gets low?

You get the idea. Perhaps something like this has been done, or maybe you can come up with a better experiment.

 

[yesimon]

This test wouldn't be useful because the problematic aspect of BMS miscalibration comes from inaccurate coulomb counting more than inaccurately measured degradation. Degradation is predictable, but inaccurate coulomb counting is really not, and dependent on uncontrollable factors like user behaviors.

Yes, the risk is the SOC "suddenly" dropping from 20% to 15% while you're driving a rural highway and were counting on that 20% to get to the next charging stop. Now you would be stranded. The risk is not really dropping from 5% to 0% because 5% has enough of a voltage drop to be accurately measured by the BMS.

Another fun thought experiment: A driver who charges to 100% every night and only uses 10% per day exclusively cycles between 90-100%. The BMS is miscalibrated in the sense that it doesn't really know the full capacity, and true degradation is unknown. Yet nobody thinks of this as a problem and Tesla doesn't say to drive down to 0% occasionally. That's because degradation is mostly predictable and you get much more advance warning that voltage is dropping faster than expected.

 

[PianoAl]

When you say "degradation," are you referring to a drop in SOC?

Good point about a drop from 5% to 0%.

I still would like to see a real-world example of how a miscalibration manifests itself.

 

[AAKEE]

When you say "degradation," are you referring to a drop in SOC?

Good point about a drop from 5% to 0%.

I still would like to see a real-world example of how a miscalibration manifests itself.

We already know this.

Theres a lot of exsmples of older model S suddenly loosing the last 10-20 miles very quick and stopping, which probably comes from a miscalculated battery capacity.

For a LFP tesla the buffer increases to compensate the non exact SOC Measurment. This mesns s larger margin to cope with the uncertain SOC level.

I know a LFP car that goes 100-0% or close to 5x each week.
He charges to 100% daily so the BMS should not be uncertain about SOC and the buffer should be low.

He got stuck with a few percent remaining when the car stopped. This should not have been a uncertain SOC level initially but probably a overestimation of the capacity.

 

[jensk2]

1. Keep the SOC between, say 40% and 70% for a thousand(?) miles.
2. Drive until the car runs out of charge.

That may show how "off" the BMS calibration gets. Does it run out of charge when the display is showing 20%? Does the car have lots of charge left even when it is displaying 0%? Does the displayed SOC suddenly change as it gets low?

You get the idea. Perhaps something like this has been done, or maybe you can come up with a better experiment.

Yes, It is a good test. It should as well reveal that appart from the BMS calculating a more true capacity, LFP have reduced 'real' capacity when always shallow cycled!

So it is not only a BMS challenge from the very, very stable Voltage at different SoC, but as well an inhomogenity in the charge levet of the electrodes. So when you charge to 100% on LFP you even the charge homogenity (towards the entire electrode area having the same 'maximum' potential.

So you actually INCREASE the total capacity by either emptying electordes fully or 'filling' them fully. Think of it as the active electrode area being smaller and smaller, when you do narrow cycles. The recovery effect was initially thought to be Anode Overhang Recovery, but is not.As it is better to age at low SoC, it is better to recuperate at 0% SoC instead of at 100%

GRID LFP batteries are wide cycled periodically to eliminate the inhomogenity. The effect is pretty dramatic, say recovery from 83% capacity to 97% capacity!

Radware Bot Manager Captcha

 

[jensk2]

Here's the experiment I'd like to see someone do (perhaps a YouTuber):

1. Keep the SOC between, say 40% and 70% for a thousand(?) miles.
2. Drive until the car runs out of charge.

That may show how "off" the BMS calibration gets. Does it run out of charge when the display is showing 20%? Does the car have lots of charge left even when it is displaying 0%? Does the displayed SOC suddenly change as it gets low?

You get the idea. Perhaps something like this has been done, or maybe you can come up with a better experiment.

To complete my reference to Anode Overhang, here is a graph showing how much Anode Overhang Recovery can be! The graph shows recovery (by resting a violated Prismatic Cell (Not LFP) first with a zillion either 6% or 12% SoC Cycles around P1==81% SoC, and the capacity Recovered when instead resting the cell at P5==9% SoC (to 'pull' the LI-Ion's pressed out inthe inactive Anode Area, where there is no coressponding Cathode area,, as the anode generally has a larger area than the cathode)

Graph stolen from: https://www.sciencedirect.com/science/article/abs/pii/S2352152X18306637

(Observe that there is even as small recoveriy when resting at P1=81%. That is likely explained by the fact that it is the charging that worst press ion's into the less accesible area )