HV Battery Died with 7 miles range left showing on Range display

[LostVector]

Electrolyte oxidation happens at very high voltage - in general above the maximum voltage lithium ion batteries use.

4.20V/cell is the 100% that Tesla use. The examples in Jeffs presentation is above 4.20V.

Of course I do not know for a fact that this very particular mode of failure is what's going on in OP's battery, but it's called out in the presentation and clearly OP's battery is defective in some way. Averages and expected behaviors do not apply.

I havent studied the electrolyte oxidation specifically but I would guess that this is a non issue for EV users.

We do not use high voltage that often in general, as daily charging is ~80-90% (4.0-4.1V) or below. Most of us do not cycle the cells to 4.20V each time.

You're assuming the battery pack is degrading in a normal way. It clearly hasn't and so isn't there a different failure mode at play? Imagine something was out of tolerance on one or a few batteries in the pack and there's a faster than expected buildup on the electrodes causing a few cells to start losing capacity quickly. It does not have to be a high charging voltage causing the issue.

The point is I do not think you can look at studies designed to measure and predict the behavior of batteries within design spec when in all likelihood something is out of spec.

The failure mode described do not mean this. It means that the cell rapidly looses capacity in a low number of cycles, and then die - dead like in not possible to charge.

I'm a little unclear on what you are arguing relative to what I said about sudden drop in voltage and shutdown by the BMS.

When EV’s stops at whatever percentage above 0% displayed it do come from the fact that SOC can not be measured when driving (or charging).

Why do you think this is the case? I was under the impression that the BMS can get a more accurate reading when the battery has rested, but there are still certainly many data points that can be collected while the battery is under load to continue updating the SoC. Do you have a different understanding of this?

The State of charge is measured by measuring the resting voltage (OCV) of the battery (-cells).

Again, rested is the best way to measure it, but I don't understand it as the *only* way.

When driving (or charging) the BMS need to calculate the estimated SOC.
The calculation depends on the estimated capacity (and also the SOC). Any fault in estimated capacity or estimated SOC before the drive ends up in a error in the resulting estimated SOC we se on the display.

I agree? But also this seems a bit tautological?

 

[EVer Hopeful]

I'm curious what was going on as it shut down? Were you going uphill?

Funny you should say that (right at the beginning of this thread) but I have had that happen in an icemobile - but that was because the pickup must have been at the front of the tank causing it to cavitate when the fuel sloshed to the back

I've heard similar (but opposite) happens in a Cessna 182 and that a forced landing from fuel starvation can go awry in the flare as the nose comes up, fuel runs to the back of the fuel cell and the engine picks back up destroying all your careful calculations

Maybe the electrons are doing something similar

 

[AAKEE]

When EV’s stops at whatever percentage above 0% displayed it do come from the fact that SOC can not be measured when driving (or charging).

The State of charge is measured by measuring the resting voltage (OCV) of the battery (-cells).

Why do you think this is the case?

I would not call it *think*…

As you put a load on a battery, the voltage droops. The voltage droop vill vary with load, cell temp and SOC (internal resistance). It will also vary with the recent load history. There are too many variables to measure the drooped voltage and calculate the SOC from this.
If you are driving, and are stopping (at a red light for a minute, or something similar) and the battery voltage clearly shows that it is different than the estimate the BMS can adjust the SOC from this but this will be a rough adjustment.


The same is valid for charging - The SOC can not be measured in a reliable way during charging as we feed the batteryu with a higher voltage than the cell voltage to create the current. So, the BMS calculates the needed amount of energy to hit the set SOC level. Looking at the BMS values during a charging session we can see that there is a value “to charge comnplete”. This value is calculated before the charging commences and is calculated by “needed change in SOC x capacity”.
If the BMS overestimates the capacity, the end result SOC will overshoot the planned/requested SOC. This as the calculated amount of energey will be more than what was needed to hit the target SOC. This is quite common in threads for people to ask why it end up higher then the set level.
Seing a overshoot after a charge (need not to be a to small charge session) indicates that the BMS overestimates the capacity.

Same thing after a drive, but the other way around: If the SOC droops after a (longer) drive, it is most probably due to a BMS overestimating the capacity. If the SOC recovers after a (longer) drive, its probably a BMS underestimate. Using Scan my tesla or similar, so the SOC can be read in better resolution makes this very clear.

I never had any gross overestimate, but my M3P had a large underestimate. I developed a way of calculating the capacity during that time, and also drove a 100-0% drive to check the capacity.
During this underestimate any longer drive made it clear that the SOC recovered after a drive (see around 10 a clock).The 100-0 drive on the picture, and arrival after 240km with 52% SOC and after several hours of sleep it showed 54%. Same thing when arriving hime after the return drive, the SOC started to recover.

When I got my Plaid, the BMS hads the battery capacity at 96kWh, but the new capacity is aorund 99.4kWh when leaving the manufaqcturing line.

So, the same thing happened here. Every longer drive had a recovery afterwards. Every. Like this:

So, it was even possible to calculate the true capacity by using *Estimated used SOC/Real used SOC x BMS estimated capacity* to find the real battery capacity. This as the end SOC is a estimated value from the estimated capacity.
Also, the better way to calculate the capacity is *delivered energy/ delta SOC*. It will be very precise as long as the the power during the drive was not to high (speeds 90-100kph/55-60mph) as the specified capacity is at a certain battery load.

Well, the BMS adjusted slowly, and after one month the BMS came to the same capacity estimate that I had done ( = 98-98.3kWh). From that day, the SOC doesnt recover after a longer drive.
Scan my Tesla shows this very precise with SOC in one or two digits, but for the purpose of this forum, teslafi graphics show this better (but not as precise).

A full charge (100%) is not estimated, as in the end of the charge the supply voltage is held constant until the current has reduced below a (low) level. To think it in an easy way, the voltage is held at 100% SOC voltage until the battery stops receiving any more current.

 

[AAKEE]

Again, rested is the best way to measure it, but I don't understand it as the *only* way.

The “stop” above 0% is not extremely uncommon. And it doesnt need to be a battery failure, it just need to be a bad capacity estimate.

Think about how it can be possible for several cars to stop above 0% if the BMS could measure the SOC well during a drive.
Tesla use 4.5% of the total capacity below 0% displayed, so there should be a good margin before the car stops in the normal case.

 

[LostVector]

The “stop” above 0% is not extremely uncommon. And it doesnt need to be a battery failure, it just need to be a bad capacity estimate.

Think about how it can be possible for several cars to stop above 0% if the BMS could measure the SOC well during a drive.
Tesla use 4.5% of the total capacity below 0% displayed, so there should be a good margin before the car stops in the normal case.

If it were a bad capacity estimate on a pack with healthy cells, why would the pack go immediately into shutdown? Wouldn't you expect the remaining estimate to drop unusually quickly first as it enters the tail end of the discharge curve, not just enter a shutdown state (especially when there is supposed to be 4.5% buffer)?

On the other hand, consider a pack with some cells experiencing premature failure ... I believe the BMS would detect an unusual and unexpected voltage drop in either individual cells or the pack, then shutdown the car quickly to prevent serious damage to any cells from overdischarging. That looks to me like what's being experienced here. Do you have a different take?

I would not call it *think*…

As you put a load on a battery, the voltage droops. The voltage droop vill vary with load, cell temp and SOC (internal resistance). It will also vary with the recent load history. There are too many variables to measure the drooped voltage and calculate the SOC from this.
If you are driving, and are stopping (at a red light for a minute, or something similar) and the battery voltage clearly shows that it is different than the estimate the BMS can adjust the SOC from this but this will be a rough adjustment.


The same is valid for charging - The SOC can not be measured in a reliable way during charging as we feed the batteryu with a higher voltage than the cell voltage to create the current. So, the BMS calculates the needed amount of energy to hit the set SOC level. Looking at the BMS values during a charging session we can see that there is a value “to charge comnplete”. This value is calculated before the charging commences and is calculated by “needed change in SOC x capacity”.
If the BMS overestimates the capacity, the end result SOC will overshoot the planned/requested SOC. This as the calculated amount of energey will be more than what was needed to hit the target SOC. This is quite common in threads for people to ask why it end up higher then the set level.
Seing a overshoot after a charge (need not to be a to small charge session) indicates that the BMS overestimates the capacity.

Same thing after a drive, but the other way around: If the SOC droops after a (longer) drive, it is most probably due to a BMS overestimating the capacity. If the SOC recovers after a (longer) drive, its probably a BMS underestimate. Using Scan my tesla or similar, so the SOC can be read in better resolution makes this very clear.

I never had any gross overestimate, but my M3P had a large underestimate. I developed a way of calculating the capacity during that time, and also drove a 100-0% drive to check the capacity.
During this underestimate any longer drive made it clear that the SOC recovered after a drive (see around 10 a clock).The 100-0 drive on the picture, and arrival after 240km with 52% SOC and after several hours of sleep it showed 54%. Same thing when arriving hime after the return drive, the SOC started to recover.
View attachment 1040954


When I got my Plaid, the BMS hads the battery capacity at 96kWh, but the new capacity is aorund 99.4kWh when leaving the manufaqcturing line.

So, the same thing happened here. Every longer drive had a recovery afterwards. Every. Like this:
View attachment 1040969


So, it was even possible to calculate the true capacity by using *Estimated used SOC/Real used SOC x BMS estimated capacity* to find the real battery capacity. This as the end SOC is a estimated value from the estimated capacity.
Also, the better way to calculate the capacity is *delivered energy/ delta SOC*. It will be very precise as long as the the power during the drive was not to high (speeds 90-100kph/55-60mph) as the specified capacity is at a certain battery load.

Well, the BMS adjusted slowly, and after one month the BMS came to the same capacity estimate that I had done ( = 98-98.3kWh). From that day, the SOC doesnt recover after a longer drive.
Scan my Tesla shows this very precise with SOC in one or two digits, but for the purpose of this forum, teslafi graphics show this better (but not as precise).
View attachment 1040978



A full charge (100%) is not estimated, as in the end of the charge the supply voltage is held constant until the current has reduced below a (low) level. To think it in an easy way, the voltage is held at 100% SOC voltage until the battery stops receiving any more current.

I think we are talking a bit past each other because although what you say is true, the BMS is still estimating the SoC based on ongoing info available to it, both while driving or resting. Saying the SoC cannot be measured reliably while driving is a matter of degrees, not a binary yes it can or no it can't. I understand why you are saying it, but to me it's inaccurate to characterize the measurements and estimations taken while driving as completely useless.

 

[flixden]

I am sure the car measures the current going in and out, and should be able to make a fairly accurate SoC calculation based on that, combined we the last known resting SoC. No?

 

[E90alex]

I am sure the car measures the current going in and out, and should be able to make a fairly accurate SoC calculation based on that, combined we the last known resting SoC. No?

I mean yes the BMS takes all that information into account. But measuring SOC is still a bit of a guessing game because of all the variables involved. Temperature differences, conversion losses, heat losses, degradation losses, cell imbalance etc etc.

It’s never going to be as straightforward and precise as measuring physical gallons of fuel.

 

[flixden]

I mean yes the BMS takes all that information into account. But measuring SOC is still a bit of a guessing game because of all the variables involved. Temperature differences, conversion losses, heat losses, degradation losses, cell imbalance etc etc.

It’s never going to be as straightforward and precise as measuring physical gallons of fuel.

I would tend to disagree.
The fuel gauges and miles left on gas cars are not all that accurate from my experience.
With precise current over time measurements, combined with idle readings, EVs will get there. Conversion losses and temperature can all be factured in.

 

[AAKEE]

I am sure the car measures the current going in and out, and should be able to make a fairly accurate SoC calculation based on that, combined we the last known resting SoC. No?

SOC is per definition a resting/ OCV Voltage.

So while driving/charging you get a voltage droop / increase so you can not measure the SOC. The drooop/increase depends on to many factors to safely being able to be calculated.
So having 3.65V OCV per cell or so at rest, the battery voltage might measure 3.0V during a fast pull or 3.8V during a regen cycle.

The exakt same regen cycle at 3.65V might cause the voltage to be 3.7 or 3.8V depending on the short term history and cell temperature.
So the variations are quite high.
After a regen a 3.65V cell might read 3.7V for a while despite having 3.65V OCV, or it might read 3.6V if it delivered power recently.
After ~15 minutes or so, these to would both read 3.65V or close, one having a reducing voltage reaching 3.65V and the other a inteasing voltage reaching 3.65V.

Dueing a drive the BMS counts the delivered energy from the battery.
If you have 100% in a 100kWh battery, and it has delivered 50kWh, the real SOC could be 48% or 51% depending on the:
-Cell temp: lower cell temp gives more losses.
-Battery load (power). Higher power means more losses, and the specified 100kWh is during a certain load at a certain cell temp.
-Degradation: the higher SOC the battery mostly is at, the higher the internal resistance will get with time.
We can not use battery load and cell temp only as the varying internal resistance also has a clear impact of the losses.

= So the BMS use the measured SOC and estimated capacity to find a nominal remaining energy from which it subtracts the delivered energy which gives us the new nominal remaining.
The new nominal remaining is divided with the estimated capacity to give a estimated SOC.
(Which might be wrong from having the wrong capacity estimate or the wrong losses calculated from load/cell temp/internal resistance).

When parking, the real SOC (or very close) is read within 15-20 minutes and the actual SOC can be read. For a precise SOC, the battery needs to be disconnected (car sleeping) for a while when the OCV creeos to the true resting voltage.

 

[ewoodrick]

I would tend to disagree.
The fuel gauges and miles left on gas cars are not all that accurate from my experience.
With precise current over time measurements, combined with idle readings, EVs will get there. Conversion losses and temperature can all be factured in.

You are half right.

Yes, the gauges used in cars are notoriously inaccurate. How can a fluid height sensor in a tank of liquid sloshing around be anywhere near correct.
But then there are fuel flow measuring devices that are really accurate. And there's even the ability to measure the fuel level when the vehicle is at stop, including if it is unlevel.

And while temperature does impact the volume of gas, it doesn't impact it as much as temperature impacts batteries.

There is one and only one way to accurately measure the capacity of a battery. Charge it to 100% and then discharge to 100%.
If that doesn't damage the battery, then what you did was measure the batteries capacity "the last time" and that's going to be different than the next charge.

Battery charge level is an ART, backup up by a lot of science.

 

[brainhouston]

SOC is per definition a resting/ OCV Voltage.

So while driving/charging you get a voltage droop / increase so you can not measure the SOC. The drooop/increase depends on to many factors to safely being able to be calculated.
So having 3.65V OCV per cell or so at rest, the battery voltage might measure 3.0V during a fast pull or 3.8V during a regen cycle.

The exakt same regen cycle at 3.65V might cause the voltage to be 3.7 or 3.8V depending on the short term history and cell temperature.
So the variations are quite high.
After a regen a 3.65V cell might read 3.7V for a while despite having 3.65V OCV, or it might read 3.6V if it delivered power recently.
After ~15 minutes or so, these to would both read 3.65V or close, one having a reducing voltage reaching 3.65V and the other a inteasing voltage reaching 3.65V.

Dueing a drive the BMS counts the delivered energy from the battery.
If you have 100% in a 100kWh battery, and it has delivered 50kWh, the real SOC could be 48% or 51% depending on the:
-Cell temp: lower cell temp gives more losses.
-Battery load (power). Higher power means more losses, and the specified 100kWh is during a certain load at a certain cell temp.
-Degradation: the higher SOC the battery mostly is at, the higher the internal resistance will get with time.
We can not use battery load and cell temp only as the varying internal resistance also has a clear impact of the losses.

= So the BMS use the measured SOC and estimated capacity to find a nominal remaining energy from which it subtracts the delivered energy which gives us the new nominal remaining.
The new nominal remaining is divided with the estimated capacity to give a estimated SOC.
(Which might be wrong from having the wrong capacity estimate or the wrong losses calculated from load/cell temp/internal resistance).

When parking, the real SOC (or very close) is read within 15-20 minutes and the actual SOC can be read. For a precise SOC, the battery needs to be disconnected (car sleeping) for a while when the OCV creeos to the true resting voltage.

This all sounds great but do u have any proof that this is how BMS works?
I'm pretty sure it can use voltages during driving/load also for its calculations
There's averages and under small loads on cells, voltage drop is insignificant.
Plus ur car stops for traffic lights, coasts sometimes, etc, plenty of times when load almost equals to OCV..
OCV is not always accurate either, i've seen cells that show good voltage on DVM but as soon as u apply load it drops to nothing n has no juice left...

 

[ewoodrick]

This all sounds great but do u have any proof that this is how BMS works?
I'm pretty sure it can use voltages during driving/load also for its calculations
There's averages and under small loads on cells, voltage drop is insignificant.
Plus ur car stops for traffic lights, coasts sometimes, etc, plenty of times when load almost equals to OCV..
OCV is not always accurate either, i've seen cells that show good voltage on DVM but as soon as u apply load it drops to nothing n has no juice left...

Dou you not know that when stopped for traffic lights that the computer is still powered up, tail lights are on, A/C is on?

When millivolts are what you are measuring to determine capacity, millivolts matter.

Plus the simple answer, WHY do you need to?

 

[brainhouston]

Dou you not know that when stopped for traffic lights that the computer is still powered up, tail lights are on, A/C is on?

Did u not know that when car is asleep computers are still powered on
Yeah yeah i know HV is off lol

When millivolts are what you are measuring to determine capacity, millivolts matter.

U do not get my point, when u have 7k cells, few amps of current will equal to few mA from each cell, its not significant enough to sag ur voltage

Plus the simple answer, WHY do you need to?

What if u set out on long trip n drive non-stop from 100-0%??
U seriously think BMS just gonna guess?
No way, that would be very poor engineering...

 

[ewoodrick]

U do not get my point, when u have 7k cells, few amps of current will equal to few mA from each cell, its not significant enough to sag ur voltage

What if u set out on long trip n drive non-stop from 100-0%??
U seriously think BMS just gonna guess?
No way, that would be very poor engineering...

I am getting your point. Even mA impact the voltage. Sure, it may not be much, but that's indeed what you are measuring with the BMS, mV.

BMS doesn't have to guess going from 100-0%, it can measure the current and make a determination. Now of course that it going to be off a little. But then again, once the batteries cool down from the drive, the numbers are going to change as well.

Just some rough numbers, the voltage drop is less than 0.5V from full to 10%. So 1% is 5mV. That's not much.
But in reality, during much of the discharge cycle, they voltage varies by a lot less than 5mV.

Try it, go get a cell and do the measurements on it. You may need to get a good VOM to measure the voltage though.

 

[AAKEE]

This all sounds great but do u have any proof that this is how BMS works?

I have three year of logs from the BMS, showing this on two cars.
Also, if one want proof, just read research about how batteries work.

When my two Teslas had a underestimate of the capacity, the SOC always ended up below the set charging level, and the resulting SOC could be calculated from the underestimation (estimated capacity/real capacity).

Overestimation of the capacity leads to the calculated SOC being to high, which cause the SOC to drop a short while after the drive.

I'm pretty sure it can use voltages during driving/load also for its calculations
There's averages and under small loads on cells, voltage drop is insignificant.
Plus ur car stops for traffic lights, coasts sometimes, etc, plenty of times when load almost equals to OCV..

You need to go back and read about what I wrote earlier about for example stopping at a stoplight. If the car is at a short park and the estimated SOC is appearently off (for exsmple reading a voltage well above the voltage that the calculated SOC should be the car can adjust it Slightly but it will not be spot on.

To be clear, even if you shut down the car at the stoplight (no load, the OCV vill be different. It takes a while for the Ocv to stabilize.

OCV is not always accurate either, i've seen cells that show good voltage on DVM but as soon as u apply load it drops to nothing n has no juice left...

Thats a bad, useless cell so…

OCV is accurate. What we mainly mean with OCV is the open vircuit voltage when it has stabilised (depending om load etc, maybe 1-2hours after the load was taken away).

We do not need to discuss the relation between SOC and OCV, its established by for exsmpke research.

 

[AAKEE]

I am getting your point. Even mA impact the voltage. Sure, it may not be much, but that's indeed what you are measuring with the BMS, mV.

BMS doesn't have to guess going from 100-0%, it can measure the current and make a determination. Now of course that it going to be off a little. But then again, once the batteries cool down from the drive, the numbers are going to change as well.

Just some rough numbers, the voltage drop is less than 0.5V from full to 10%. So 1% is 5mV. That's not much.
But in reality, during much of the discharge cycle, they voltage varies by a lot less than 5mV.

View attachment 1042075

Try it, go get a cell and do the measurements on it. You may need to get a good VOM to measure the voltage though.

That is a LFP / LiFePo4-battery with a flat voltage curve.


OCV tested on a Pana NCA cell.

 

[ewoodrick]

That is a LFP / LiFePo4-battery with a flat voltage curve.


OCV tested on a Pana NCA cell.

Okay, a little more slope, but still mV are being measured.
Just swagging, 0.2V for 20% in the middle.
10mV per percent.

 

[AAKEE]

I'm pretty sure it can use voltages during driving/load also for its calculations
There's averages and under small loads on cells, voltage drop is insignificant.

This is a recent drive. From 100-0% displayed we have more or less 1.0V, so 1000mV = 100%, and 10mV (0,01V) equals 1%.
The drive started at 69% SOC and ended at 15%, so 54% used.
The drive started at 3.93V and ended at 3.38V so 0.55V so we can see that the 1% = 0.01V is a good approx.

You can see the voltage going up and down around 0.05 V alla the time, so thata 5% SOC if you try to measure it.

There also is one 90mV which equals 9% and one 130mV = 13% step, so not easy to measure during a drive.

For the going slow, after another longer drive the car was parked at the grocery shop for a shorter while before going home.

The first screenshot is before the shop, but when driving 50kph or less in town, the other three below is after the shop, still in town driving 50kph or less.

Between the frist two, the car estimates the SOC to be the same but the voltage differs 110mV, so around 11% SOC if you measure it.

The lower two, still 16% SOC but the voltage is in between the first two.

 

[QtownTaxiGreg]

Slam dunk, final buzzer?

Quick question. How much of this is applicable to the "330 mile" battery in 2024 MYLR when it is used for maybe 1/4 to 1/2 a full charge most days, with little to no advance notice of longer trips?

There's a 300kW big boy charger going to be installed in town soon enough. I imagine the half pint Tesla SC by the airport will need to be upgraded from 120 to 250kW but that may take a bit longer as it gets little enough use as it is.

Point being a 10 minute suck on one of those badboys would take it from 30% to 80% if I needed to make tracks. Not very different from filling up with diesel except with diesel that's the filling done for the day.

 

[brainhouston]

View attachment 1042106

i think we both saying the same thing
yes voltage fluctuates but if u do a straight line fitting in above graph you'll get nice straight downslope for average
i'm sure BMS uses everything it can for estimation, i just don't think it ignores voltage readings during drive