Wiki Sudden Loss Of Range With 2019.16.x Software

[sorka]

FYI, Tesla algorhythm has always changed the way that it comes up with rated range. Rated range=ideal kwh remaining/Wh/mi constant. Mine started out as 300Wh/mi, and now it's at 265 according to the BMS.

Um no

 

[Bmod]

Here is what I learned from briefly reading the patent without analyzing it too much. In overall the BMS has a preconfigured set of parameters that the BMS uses to set a fault flag that modifies the charge parameters (possibly e.g. the max cell voltage).

When I talked earlier about detecting imbalance of cells I did not mean the actual cell voltages as likes someone said that OCV is a poor prediction of SoC. Rather when talking about imbalance of cells we need to take into account several possible algorithms.

The figures 3-8 in the patent demonstrate the trigger conditions. I am not an expert of lithium chemistry but someone who has that bacground can propably analyze the causes for each of these different conditions. My bet is that after the recent battery fires they just made the detection limits way more tight.

Even though the fault flag is set the patent lists several false positives like: Note that with respect to decreasing voltage, this method monitors for false trips due, for example, to (i) step change from one CC level or CP level to another, (ii) reduction in charge rate due to grid power or charger power limits in the presence of an HVAC load for example, (iii) external heating of the battery causing a reduction in impedance, or (iv) reduction in impedance at a rate which causes the loaded voltage to fall more quickly than the OCV rises.

My second bet is that due to the recent fires the newly imposed detection limits were not really tested and hence most likely yielding a fair amount of false positives. Like someone here said already "we are likely to see restored ranges in the future". Regardless of the outcome the Tesla customer communications is a real bummer - to me its like any typical engineering company with lack of ability to communicate whats important for customers...

Here is a repost of the link to the patent for the curious minds: https://patentimages.storage.googleapis.com/d4/c7/2b/48b50249e2c13e/US8866444.pdf

And below are the conditions we should start looking for to see if its a false positive or a real concern.

Methods 1 and 2 above (i.e., decreasing Voltage during charge and increasing current during CV charge) can be mea
Sured during any charge cycle that begins with V<V, and deltaT<(deltaT), where deltaT is the maximum difference
in temperature between the cells throughout the battery pack and the ambient temperature and (deltaT), is the maximum acceptable difference in temperature that would affect a change in Voltage or current during charging, this value being
determined through testing and contained in a look-up table. The battery-management system verifies that the charge cur
rent has not decreased or been interrupted, and that no addi tional loads have been introduced (i.e. HVAC) before indicat
ing a hazardous condition.

Methods 3 through 7 use time or capacity measurement. For methods 3 and 4, a time counter can be used to determine
if the battery has been charging longer than the time indicated in a look-up table for a pack of that capacity and state of
charge. This look-up table may be compiled from pack data, as well as from resistance measurements and cycling data
from cells at end-of life. These measurements also require that deltaT<(deltaT).

For method 5, the self-discharge rate of the pack can be measured and compared with the self
discharge rate for a cell of that capacity without an internal cell short present (from a look-up table). This can be mea
Sured using Voltage measurements over time, or, if a bleed circuit is present, it can be measured by determining the
amount of Ah passed through the bleed circuit to balance the series blocks of cells. These methods take into consideration
shallow and full charges, as well as SOC inaccuracy.

Method 6 (charge efficiency) is similar to method 5 in that it measures the amount of self-discharge. This method also
requires that any temperature, rate, and cutoff Voltage differ ences between charge and discharge are accounted for. If an
internal-cell short is present, the charge capacity will increase (some current is going through the short) and if the short is
maintained the discharge capacity will decrease relative to a cell without a short. The ratio of the discharge to the charge
capacity should be 1 for a fully reversible cell without shorts or side reactions. This value is typically ~99.97 or higher for
new cells without shorts. For a cell with an internal short, this value may be ~70% or lower. A low-end limit value is deter
mined through testing for the cell in use and this value will be included in a look-up table for the battery management sys
tem.

Method 7 may be the most simple to implement for some applications, especially in a large battery pack Such as the
type of pack that might be used in an electric vehicle. This method measures the total charge passed to the pack, module,
or parallel brick. If the total charge passed exceeds the expected charge capacity from the initial to final state of
charge (SOC), then the additional charge may have been dissipated through internal shorts. In a vehicle battery, the
capacity is continuously monitored in order to estimate remaining range. The calculated Ah capacity (CAC) is moni
tored as the packages, enabling this method to be used throughout the life of the battery pack. Say for example that a
100 Ah battery pack is assembled and it is at 20% SOC initially, or 20 Ah of energy is available in the pack. This
example pack contains cells with internal shorts and takes 130 Ah (an additional 110 Ah) to reach full pack voltage. This
means that 30 Ah were dissipated through internal cell shorts and caused the pack to heat up, potentially creating a hazard
ous condition. By comparing the total charge passed to the expected value, this method may be used to interrupt charge,
potentially before a hazardous condition can occur. Method 7 is doubly useful in that it can also be used as another layer of
protection against pack overcharge, a highly hazardous con dition should it occur.

Method 8 involves a real-time measurement of the impedance of parallel bricks of cells, and potentially also series
strings of cells. This measurement is continuously made during pack operation to estimate available power using a com
bination of Voltage and current measurements. During charge, these measurements are typically not made. However,
Some embodiments of this invention propose use of a continuous estimation of battery impedance to detect an internal
short. For a given SOC, the expected impedance can be kept in a look-up table. If the impedance at that SOC is lower than
the threshold value, it may indicate that an internal cell short has formed. This would capture the case of a decreasing
US 8,866,444 B2 Voltage during CC or CP charge as well as an increase in current during CV charge. As indicated above, any additional load on the battery pack or change in temperature is properly accounted for.

Method 9 involves the direct measurement of a resultant increase in temperature due to an internal-cell short. The cell,
brick, and coolant temperature (refrigerant, water, air or another working fluid) is monitored continuously. If it
increases intemperature more rapidly and/or to a higher value than is expected for that charge condition, a hazardous con
dition is indicated and the charge is interrupted.


View attachment 428297

Lots of info here. Thank you for doing the research to help us out to understand this better.

 

[Zodiacchris]

My 70D has just started a new software update. Dare I hope? They’ve had 4 weeks to work out what to do about this...

 

[lightningltd]

My 70D has just started a new software update. Dare I hope? They’ve had 4 weeks to work out what to do about this...

One can only hope. Just installed it on my 2013 S85. Will supercharge after work and see.
I HOPE they did not push this out to the entire fleet just to update sketchpad! (2019.24.1.1d133ad)

 

[neroden]

It's been a long thread, but I think at one time we believed you must supercharge after the infamous 2019.16.x update to trigger the range loss.

Hmmm. My other car, the used 2013 I just bought, was on 2019.16.x before I bought it and I Supercharged repeatedly bringing it home from Toledo. I didn't see any immediate loss after doing that. It now may be down 5 miles but it's hard to tell because I know the reported 90% range does fluctuate a little with the weather which has been changing massively (probably due to ambient battery temperature) -- I am not going to charge to 100% just to check that number, since I know that's bad for the battery.

 

[neroden]

Low PHEV DOD's made more sense, because why chance cycle-life when you can run a motor? Taycan supposedly uses less of the available 90KWh, as well. WK057 can correct, but I think the MS "85" packs would load up to around 73, of their total 81-82KWh, per his discoveries. That's ~90%. This isn't to say measured IR isn't a function of other things, like random manufacturing, or super-charging, or chronic cell imbalance. Yup, we're speculating.

I really do think it's affecting some cars and not others. I couldn't tell you why (random manufacturing variance, amount of supercharging, ambient temperatures, times charged to 100%, who knows).

(P.S. Other possibilities include frequent high acceleration, infrequent high acceleration, high use of regenerative braking, use of regenerative braking in the cold, Supercharging in the cold, WHO KNOWS)

 

[DJRas]

You are seeing what I am seeing. I supercharge often and HAD a 3% range loss at 134,000 miles. Now it is MUCH MUCH more loss:
2013 S85 Battery revision B
New: 265 miles Rated Range (100%)
Pre update: 256 Miles Rated Range (96.60% from new) (3.4% range loss from new)
Post Update (suddenly): 217 Rated Miles (84.77% from pre update)(15.23% loss from pre update) (81.88% from new) (18.11% loss from new)

The fact that it is affecting some with the same part numbers and charging habits and not others is looking to me like a clear defect in materials or workmanship.

Can you get the data from your unaffected friend?

Yes, i have many datasets from his car.
I can't figure what the trigger was.

 

[neroden]

Not even a blurb in there on safety? Basically they're just telling you that they're capping your range to make the battery last longer. Charging in a more narrow range of capacity always makes the battery last longer and Tesla advises charging to 100% very often otherwise your battery would degrade.

But that was ALWAYS the car owners choice before. If they want to degrade their battery fast by charging high, it's their choice. The situation might be that they need that range and are willing to let the battery degrade, and then either by a new battery or upgrade their car.

If what the foreman said was true, then why not cap everyone's battery and why not do it from the very beginning. It doesn't track. The public statements from tesla about the battery update preventing combustion makes far more sense, but if true warrants a recall. It's insane to try and bandaid a battery battery that might catch fire with a software patch.

Not really. All batteries can and will catch fire if you do the "wrong" things to them, and most of them won't catch fire if you do the "right" things to them. Yes, there should be a recall, but a software fix may be totally valid, if what's happening is that the previous software was not correctly identifying normal degradation and the new software is.

 

[neroden]

Simple answer is no - not all batteries with similar type numbers or in same serial number range are affected.

Based on my own experience and knowing the backgrouns being preventing fires I have a theory:

In case there is significant imbalance between cells the maximum charge voltage is reduced until balance is acquired. This way there is less risk for an overvoltage for a single cell or set of parallel cells. When balance is acquired then maximum charge voltage is restored.

The overvoltage situation can happen for several reasons, Li plating being one of those. Others are for example cell damage and charging habits - or like in my case swapping a bad module when reconditioning a battery.

Just speculation but educated speculation...

Soooo, how does the balance get restored?

 

[neroden]

The figures 3-8 in the patent demonstrate the trigger conditions. I am not an expert of lithium chemistry but someone who has that bacground can propably analyze the causes for each of these different conditions. My bet is that after the recent battery fires they just made the detection limits way more tight.

Even though the fault flag is set the patent lists several false positives like: Note that with respect to decreasing voltage, this method monitors for false trips due, for example, to (i) step change from one CC level or CP level to another, (ii) reduction in charge rate due to grid power or charger power limits in the presence of an HVAC load for example, (iii) external heating of the battery causing a reduction in impedance, or (iv) reduction in impedance at a rate which causes the loaded voltage to fall more quickly than the OCV rises.

My second bet is that due to the recent fires the newly imposed detection limits were not really tested and hence most likely yielding a fair amount of false positives. Like someone here said already "we are likely to see restored ranges in the future". Regardless of the outcome the Tesla customer communications is a real bummer - to me its like any typical engineering company with lack of ability to communicate whats important for customers...

Thanks. That sounds like an excellent summary.

 

[sorka]

Not really. All batteries can and will catch fire if you do the "wrong" things to them, and most of them won't catch fire if you do the "right" things to them. Yes, there should be a recall, but a software fix may be totally valid, if what's happening is that the previous software was not correctly identifying normal degradation and the new software is.

No battery properly maintained and properly charge will catch fire if you charge it to 4.2 volts. Any battery that is in danger of catching fire when charged to 4.2 volts was either defective from the start or improperly charged and damaged by the BMS.

 

[PetriKarj]

Soooo, how does the balance get restored?

A good question - lets assume its because of li plating. I am not a specialist of the topic but to my understanding there is a two step strategy in that case:

1) avoid charging in hot and with high voltages which would cause the situation to get worse
2) recharge with high current pulses (regen?) that will melt the plating back (see attached article for more info)

Researchers “heal” destructive dendrite growth in lithium-metal batteries

It would be interesting to know if regen is not limited (particularly when cold and partially charged) for the affected vehicles to confirm that Tesla indeed is implementing dendrite healing protocol.

Anyhow the flags causing charging restrictions can be triggered also by other reasons than lithium dendrite accumulation.

 

[lightningltd]

My 70D has just started a new software update. Dare I hope? They’ve had 4 weeks to work out what to do about this...

Expect this update to suddenly stop or be recalled. Found and called in a serious safety issue with this update. Safety issue aside, it did not improve charging or range. Actually it seems to have gotten worse I’m charging I was down to 48 kW

 

[Droschke]

I really do think it's affecting some cars and not others. I couldn't tell you why (random manufacturing variance, amount of supercharging, ambient temperatures, times charged to 100%, who knows).

(P.S. Other possibilities include frequent high acceleration, infrequent high acceleration, high use of regenerative braking, use of regenerative braking in the cold, Supercharging in the cold, WHO KNOWS)

Very good list @neroden , let's see how it applies to me car (4 years old, 43k miles), which has lost +30 miles:

amount of supercharging

Yes. I do supercharge. Infact, I read somewhere years ago when the crowd-sourcing of the battery degradation data started that the cars with the most frequent supercharging were showing lower degradation (taxi's for instance) and also read the reason had something to do with the lower duration of heat buildup during charging (it's not the heat during the charging sessions, but how long the battery is exposed to the heat).

ambient temperatures

Very hot summers, very cold winters

times charged to 100%

Less than 5 times in 4 years. Always drove right after.

frequent high acceleration

No.

infrequent high acceleration

Yes.

high use of regenerative braking

Yes.

use of regenerative braking in the cold

Yes.

Supercharging in the cold

Yes.

 

[TJtv]

Expect this update to suddenly stop or be recalled. Found and called in a serious safety issue with this update. Safety issue aside, it did not improve charging or range. Actually it seems to have gotten worse I’m charging I was down to 48 kW

Can you elaborate?

 

[fbitz777]

1) avoid charging in hot and with high voltages which would cause the situation to get worse
2) recharge with high current pulses (regen?) that will melt the plating back (see attached article for more info)

Researchers “heal” destructive dendrite growth in lithium-metal batteries

It would be interesting to know if regen is not limited (particularly when cold and partially charged) for the affected vehicles to confirm that Tesla indeed is implementing dendrite healing protocol.

Anyhow the flags causing charging restrictions can be triggered also by other reasons than lithium dendrite accumulation.

Interesting as my car Regen's works at 100 percent charge (which it did not before update)

 

[bhzmark]

a software fix may be totally valid, if what's happening is that the previous software was not correctly identifying normal degradation and the new software is.

Exactly the issue. And part of the confusion is what constitutes normal degradation. The fact that a battery physically changes over time depending on use history and environment or even depending on production variations — all those changes are tracked by the BMS and factored into how the BMS (and the allowable peak and sustained power delivery algorithms) should solve for various and sometimes competing objectives such as maximizing range and performance and safety issues and longevity etc. I’m sure the calculus is constantly changing and being refined. Sometimes it may be too conservative, and sometimes too risky and later versions updated to refine that. This is the main reason that I will never not install an update.

I think the better way to think about this is that the bottom line outputs of performance, charge rate, and range are likely to degrade over time. Just accept that and don’t worry about how it gets there. Just ask: How much is the total degradation? Is it beyond the tolerances for the given amount of time? Use the Model 3 warranty as a benchmark: 70% over 100k miles. Anything worse than that is an outlier that should justify a repair or replacement by Tesla. Anything within that tolerance is acceptable — even if you think you know better than the BMS or think the BMS should allow some additional outputs that it has determined need to be reduced.

It could also be that without software intervention the safety profile of a battery could degrade over time, but with software intervention the safety and MTBF levels could be maintained, but almost certainly at the expense of some performance, charge rate, or range outputs. They could do it the other way: maintain fixed performance and range output while increasing risk of adverse safety or failure events but the obvious right trade off is preserve the most important safety and failure risks while managing the other variables to trend down as necessary to preserve the right adverse event risk profile. Whether it is described as coming about through software or hardware is a false distinction: the software monitors the hardware states and manages the other variables accordingly.

 

[Oldschool496]

No cheating, you need to read the whole thread, all 67 pages

I tried, I really did, I got to like 43 and was tired.

Too Funny

 

[Icer]

For some, it could be LI contaminates?
Metallic = bad.
Q: Can BMS identify this?

/OT:
request line: i very much would like to see Any conversation
WK057, PetriKarj, ingineer et al
Specifically:
How or what is Tesla able to observe,
what problems can they identify, what can they not identify
*
Can I observe these things in a hacked rooted never to be updated car?

Further/OT:
S85, upgraded to a P100, (Green, moonroof, tan leather)
AP1 running V8
Black box (in yellow like 57's car I wished I had the cash to buy)
Rooted, Not on the network
(Cars name would be FittySevin, BTW)
*
Can this be safely run?
What would I need to monitor and watch to know the battery health
and especially safety.

Now we return to our r3gular programming
Thank you
/OT

PS
Metallic = Bad
not to be confused with Metallica = Good

 

[Stere0graph]

It should be clear to the neutral reader that there have been posts attempting to deflect the real picture, and call it as it really is.

Tesla are behaving dishonourably with a client base they consider End of life –

Reading through the thread, members have articulated credible science as to the cause of why certain owners have been targeted.
Assuming it’s for safety – then that implies said battery is faulty and should be replaced or repaired under warranty – period.
Assuming its to prolong the life of the battery due the habits of the owner fast charging frequently – then the option to reduce range should be reserved for the owner to make – NOT Tesla’s.

It could be argued that by accepting the update you are in effect agreeing to whatever change Tesla seems fit – If that’s the case, Release notes should be made clear as to the fixes, features and implications of said update.

The cynic in me says Tesla may have targeted 85Kwh owners who predominately use their SC in an effort to curtail their use – this is to make way for the wave of model 3 owners and to alleviate the perceived oncoming congestion.

In any event I, amongst the many owners that have been effected simply want our 30 miles of range back, either by downgrading/upgrading to a release which does not exhibit the issue or by way of battery repair or replacement .