Excellent explanation. One thing I’ll start doing right away is not charging it everyday.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Range Loss Over Time, What Can Be Expected, Efficiency, How to Maintain Battery Health
- Thread starter KK_RedM3
- Start date
You could install a DC fast charger at home, provided you are willing to pay enough. First thing is to get 3 phase electricity to your home (most homes only have split phase and not 3 phase). I did ask PG&E about getting 3 phase and they gave me a rough estimate of "high 5 figures to low 6 figures" to install a new 3 phase transformer and run new service lines from it. The equipment that converts 3 phase electricity to high voltage DC to push into your car's battery isn't cheap either, and I don't see much of a point in doing this at home because it's not good for the battery in the first place to be doing it all the time. But is it possible? Absolutely. If you're willing to pay the proper people to install the service and equipment, you can do it.This is not possible.
You can have a Level 2 charger installed in your garage, but not a supercharger
That's actually not necessary if your goal is to just get the car to take readings at different charge levels. I use the "scheduled departure" function and the car generally doesn't start charging until just before I leave the next day. So whatever SoC I happen to pull into the garage with is where it takes the reading. I have my max set to 60% and most of the time, I arrive back with between 45-55%, but sometimes I'll drive further and arrive back with SoCs in the 10-40% range. The car gets to take a reading at those levels, and then it charges back up the next morning. At least once every 3-4 months, I'll push it to 100%, generally just before driving at least 30 miles, to allow the car to recalibrate.
Jeremy3292
Active Member
Then you have the opposite problem where your car never takes any readings at 70%, 80%, 90%, or 100%.
Not really. There's a mall nearby with some free chargers, as well as free chargers at the airport. If I park there, I'll set it to 90%. When I get back home, it'll generally be in the 70-80% range. It's not too often, but the car does get to take readings above 80% occasionally. It's not like it needs to do it every week in order to get a good estimate. The only point where I've never let my car get a reading is at 100%. I try not to let it ever sit above 90% for more than an hour.
I dont think any researcher will hear us
Of course the degradation will increase the day scenario 2 increases the charge to 100%. The probable possibility is that the *square root of time* and the time before with 50% charge have caused the calendar aging overall to decline, 100% might not bite as hard as if the battery was brand new. But we dont know…
The Solid Electrolyte Interphase(SEI) build up, which is the main reason for calendar aging also act as a protection that reduces degradation by time will be there when the charging schedule increases to 100% SOC.
You say: "Some people charge their car every single day to 80% or 90% regardless of how much they drove each day. So then the car never sleeps at a different state of charge so it only takes “readings” from the upper end of charge. This can cause the displayed rated miles to be inaccurate."
What I never understand is if "be inaccurate" is always less than real or it can be "be inaccurate" showing MORE rated range than real.
So performing all the steps to "Re-calibrate " the result is always a better showed rated/projectet range or it can be worse too
Jeremy3292
Active Member
A good question - my understanding is that it would not bring your rated miles down, at least I have never seen that on this forum (maybe others can chime in if they have). But remember this is just the "displayed rated miles" or what the car thinks it has "left in the tank." Your battery pack capacity in reality "is what it is" and BMS does not change reality. What it does is simply try to take the actual remaining battery capacity and calibrate itself to show the end user "how much is available for use" in the form of rated miles. So for example if the BMS isn't calibrated and it shows you 300 rated miles remaining but in reality it should be 315 rated miles when calibrated properly, you still physically have those extra 15 rated miles available to drive/use. The BMS just isn't showing you them bc of improper calibration.
Another reason why I only use % and not the rated miles as they are just an EPA constant and not how you personally drive.
Fair and great points as always - I appreciate the chemistry difference call out there too. I guess from a math perspective, using the square root principal - if you have two situations where the first year is either 1.5% or 5% degradation due to SoC differences (per original paper) - at 5 years you're at 3.4% vs 11.2%, and at 10 years you're at 4.7% vs 15.8% from aging effects alone*. That is very interesting..
I am aware of the "initial degradation" vs long term "accepted science" for calendar degradation of lithium batteries (higher initial, lower later); do we know for sure that applies to the specific chemistries that Teslas use? .. have you found any sources showing 2 years or more of data analysis to confirm this? I struggle a little with extrapolating a 6-10 year horizon off of 9.6 months of data.
*I'd also be curious as to some analysis of the inflection points between these various aging mechanics. As an analogy - if you look at all of the failure root causes for consumer grade electronics at the transistor, chip, and electrical level -- a 30-40 year old computer should never work just simply due to additive nature of all of the failure rates for each of the components; but there are thousands of examples of 8-bit and 16-bit computers from the 1970s and 1980s that work flawlessly without any modifications today. (An occasional hobby of mine
). I'm curious if lithium batteries act this way - whether there is a 'sea of stability' at some old age point (i.e. the square root math holds long term) or whether they run off a cliff in usefulness (such as the ever increasing resistance over time making them economically unviable). I will do some searching myself,
Thank you!
P.S. also Thanks for making the math easy
Thank you.
I'm totally aware/conscious about the fact that the capacity is the same (after and before a process of calibration of BMS).
What I'm curious of, is the fact that a calibrabration process can lower the range ( prediction of km left or kWh "left in the tank").
I mean... how Tesla can display to an user a Higher capacity (miles or km) if the BMS calibration is OFF?? it's a security/safety issue.
Conversely, I can tolerate(assume) that Tesla is in a conservative side, giving you LESS mileage to count on ...(knowing maybe there is MORE juice to use).
BTW, my battery seems calibrated good (or high at this point of the discussion) because every time I let it rest under 20% it looses 1 to 2 kWh of capacity (I mean Scan My Tesla Nominal full pack).
I usually charge from 20-25-30% to 50-55-60% every 2 or 3 days.
NFP is always the same but when it rest/sleep under 20% it looses capacity. So I don't like to let it go down.
Usually I regain in back (not fully back) in 2 or three weeks of normal routine (25% to 55%).
My car reached many 90% charges (one time every 2 months I do a very long trip using only supercharger) but I always use the car immediately to discharge it for the next charge for the next stage of the itinerary. My car reached many times low SOC's too ( 3 o 4%) but usually i'm in a trip so it's only because I need to charge during the trip and it doesn't sleeps at low SOC.
Jeremy3292
Active Member
Your battery definitely needs to see the upper range SOC also. There is no harm it charging it to 90% once every month and letting it sit overnight. Also I think Tesla doesn't recommend leaving it under 20% for very long as you risk discharging the 12v battery too much. Letting the BMS "see" SOC between 20% and 90% is sufficient.
Well, the BMS probably is set to make a as good estimate as possible, but still they do not want to get you stuck a few miles/kilometers short of the next charger.
My charging schedule seem to set my BMS off on the overestimating side.
I had an increase in NFP during the autumn, and it went up to 81.4kWh of 82.1 new pack size, after one year / 31.000km, which doesnt seem correct.(should have some 2.5% degradation).
A fullcharge from 17%, long drive and a another drive with 5% arrival + sleep at 5-3% set the NFP back to 80.4kWh.
I still think the NFP seems a bit on the high side, but the BMS at least did get the opportunity to see the battery high and low sides.
I'm not sure why it would matter. The BMS might estimate wrong at the high end, but even if the estimate is wrong by say 50-60 miles on the high end, wouldn't the error decrease as you get closer to the bottom? As an example, an ICE car might say you have 400 miles on a nearly full tank, because it doesn't know exactly how "full" the tank is or how fast you're going to drive. But by the time you get down to 1 gallon left in the tank, it should have a pretty good idea of how many miles you have to go until it's empty. Even a 20% error at this point is at most 4-8 miles.
A low estimate probably do not matter, as you say.
My described trip was a few days back, and I did install the 2021.44.25.2 when away.
Om the trip back I hadnt learned the new menues and I did plan to artive with low SOC at home to get a new BMS estimate.
My mental arithmetic told me the SOC should be high enough, but not wit a high margin. I started a navigation to my home, it didnt choose any SuC at the SuC I passed 50km from home, but I couldnt see the planned SOC at arrival home. I decided to go by the navigation as it always have bern good before at SOC @ the destination.
So I drove by the SuC, soon after I did get a ”maximum 114km/h to reach the destination. There was new messages with declining max speed and soon it sad ”charging needed to reach the destination”
I turned around and charged at that SuC.
I think the sudden change was due to the BMS finding that the remaining capacity was less than expected.
I did have this in mind since the first day I started to think that the BMS might overestimate the capacity, so no chock for me. I do not see a big issue with this, just writing it as information.
Kimmi
Member
To start with, you should ignore what any third party app says about your range (because tesla will) and calculate your battery capacity using the information on the car screen:
Calculating Your Battery's Estimated Capacity Using the Car's Energy Screen
Re posting here and making this a sticky post so it doesnt get lost. @AlanSubie4Life originally posted: =========================================== (originally posted by @AlanSubie4Life. Original thread this appeared in can be found here: ( 2021 model Y scan my Tesla battery size)...
teslamotorsclub.com
If People on the forum say "no, thats not normal", then what action will you take at that point? It cant be to contact tesla, because unless the car has a bunch of error messages about the battery, or is at 70% degradation before the 8 year 120k mile threshhold, they will (politely or less politely) tell you "its within spec". So, what action will you be taking if people think your capacity (as measured by the cars screen, not stats or any other third party app) is excessive for the miles (other than being upset, I mean)?
Sometimes people get mad when I ask that question, but there really isnt any other action to take beside what you likely are already doing, unless the car is at the warranty threshholds.
EDIT:... thats not quite true... you could sell the car and buy another model 3, or a different tesla, or a different brand. Thats pretty much the only actionable thing other than tracking it for information purposes.
The loss is only in the BMS and not real capacity. If you see a loss by going below 20% that should probably mean that the BMS find a error of overestimating the capacity and ajust this by lowering the capacity.
The battety itself will not loose more capacity by sleeping below 20%, its the opposite: it will loose slightly less by sleeping at low SOC.
I did have a NFP of about 80.5 during a longer period this autumn and during this time it kept 80.5 even if sleeping with 7 or 10% SOC.
Lately(one month ago) the NFR started to climb and settled at 81.4kWh which isnt really probable after one year and 31.000km. So we can believe that 81.4 is an overestimate and that the real capacity is lower(my calc say it should be in the ball park of 79.5 to 80kWh, but this is by using data from earlier chemistry( model S style 18650, which probably is more or less the same as the 2170 gen1 but not my battery which is 2170L, higher capacity).
So My BMS tend to overestimate the capacity because of my charging schedule. I did charge full before x-mas and that lowered the NFP to 81.2, but the nominal full said 81.3kWh.
On the return trip I planned to arrive with as low SOC as i could, which resulted in a onvernight SOC of 5 to 3%. (parked at 5%, and it said 3% after the night).
This was probalby due to the BMS finding the "real SOC" by a OCV measurement at about 3% SOC. After this my NFP has been stable at 80.4kWh.
I assess that the BMS is on track right now and that the 2170L probably is slightly less sensitive to calendar aging than the reference data I used. So, my real capacity probably in very close to 80.4, and I guess a new "calibration" or a very low nighly SOC wouldnt change my NFP with any noticeble number.
I do not really care about trying to keep the BMS very up to date, and I really do not care if it overestimate the capacity slightly either.
If a low SOC put you down on the NFP, then there is a high probablility the NFP is about right at that moment.
OCV = open circuit voltage.
The "real SOC" is estimated by disconnecting a battery from all loads, waiting until the voltage stabilize after the disconnect and than read the voltage.
The real SOC is read as a voltage. 4.20V = 100%, and 2.50 = 0%(real SOC).
During a drive the car reads the used kWh and guestimate the SOC by calculating used kWh compared to the calculated capacity, as the OCV isnt available.
BMS calibrating is the same thing as providing the BMS with the OCV at different voltages. The higher the difference between the votlages the better presision, as getting 20% SOC and 80% SOC forces the BMS to guesstimate the below 20% and above 80%.
Considering that the BMS can be on the OVERESITMATE side ( an eventuality that i never considered but we have to consider now ) this make me more unhappy because my battery is in real at 75 kWh capacity (NFP reads at 15%) and overextimated to around 77 kWh when in the ~30-~60% range.
And my car is the same as yours (my car is younger than yours too and with almost half of km on odometer).
So why my M3 Performance has 5 kWh less than your? It's not a real question ... it's only what I ask to myself since day one of my new Tesla Model 3 Performance).
Fair and great points as always - I appreciate the chemistry difference call out there too. I guess from a math perspective, using the square root principal - if you have two situations where the first year is either 1.5% or 5% degradation due to SoC differences (per original paper) - at 5 years you're at 3.4% vs 11.2%, and at 10 years you're at 4.7% vs 15.8% from aging effects alone*. That is very interesting..
I am aware of the "initial degradation" vs long term "accepted science" for calendar degradation of lithium batteries (higher initial, lower later); do we know for sure that applies to the specific chemistries that Teslas use? .. have you found any sources showing 2 years or more of data analysis to confirm this? I struggle a little with extrapolating a 6-10 year horizon off of 9.6 months of data.
*I'd also be curious as to some analysis of the inflection points between these various aging mechanics. As an analogy - if you look at all of the failure root causes for consumer grade electronics at the transistor, chip, and electrical level -- a 30-40 year old computer should never work just simply due to additive nature of all of the failure rates for each of the components; but there are thousands of examples of 8-bit and 16-bit computers from the 1970s and 1980s that work flawlessly without any modifications today. (An occasional hobby of mine
I will do some searching myself,
Thank you!
P.S. also Thanks for making the math easy
For the bolded(by me) part:
Well, yes we do.
There is a general accepted principle of *square root of time* as the formula for calendar aging. I have seen some research reports that try to find other formulas for a more exact estimation. For us Tesladrivers, the square root of time is good enough. We in general have to little precise data anyway.
And some of the research about this implement a straight line despite the test data they have actually show a *square root of time* line
Example: This report try to imply the degradation actually is a straight line. Their real data stop before any straight line so their conclusions is not really valid in my eyes. ( source: https://www.mdpi.com/2313-0105/7/2/28 )
Here is actual test data, 18650 NCA cells. Do not remebmer if it says Panasonic but it fore sure is cylindrical cells with NCA-chemistry.
660 days of tests(almost two years), we can see both the temperature dependent degradation as well as the SOC dependent degradation. (
3.45V = about 10-15% or so.
3.70 = about 45-50%
4.10 = about 90-95%
If I use the first clearly distinguable point on the 10C 3.7V graph and try to calculate the end point(660 days), I get 0.833% loss at day 76. using the squere root formula I end up at 2.45% degradation at 660 days and the real endpoint show about 2.2% degratation.
Of course a picture like this is har to use without fults.I did use pixel count to get as good values as possible.
For our use, the square root formula is fine. We have not the best data on our cells, as we most often have newer cells in the cars than the newest research report have. For example, the 2170L came on the market one year ago, and it probably takes one year for researches to get the hand on some of these cells and then it take another year to do a research and finalize the report.
For that we need to go back on the cars history. Some here says theres a battery lottery, but s far as research is concerned, there seem to be a very little lottery part to all of the research. There is variations, in many reports the variation is marked on the charts. Mostly maxmimum one percent or so up and down for individual cells. Between the reports( = not the same batch, or even the same cell brands or cell type) the variation is qiute small. There is no report showing 5 or 10% loss in one year if the cells had a SOC below 55% if the temperature was reasonable. One test I saw killed the cells at 60 degrees in six weeks, but constantly supplying energy so the cells stayed at 100% SOC. Not supplying voltage, the self drain lowered the SOC and helped the cell stay alive much longer. But there is not research report that really differ from the picture. Different methods make cause small differences etc, and the conclusions somtimes is really funny to read, when they made a mistake they did not realize. But the raw dtata is very clear in the reports.
What range did your car show at 100% SOC when new?
When did you get SMT and did read the first NFP?
When was your car built, and when did you get it( you are the first owner?)
Describe your charging schedule, from day one until now. How much driven daily, nighlt SOC etc. Also describe the average temperatue, for winter, summer spring and autumn. Is the car outside or inside a garage ?
OK Excellent data and that helps me a ton! Thank you!For the bolded(by me) part:
Well, yes we do.
There is a general accepted principle of *square root of time* as the formula for calendar aging. I have seen some research reports that try to find other formulas for a more exact estimation. For us Tesladrivers, the square root of time is good enough. We in general have to little precise data anyway.
And some of the research about this implement a straight line despite the test data they have actually show a *square root of time* line
Example: This report try to imply the degradation actually is a straight line. Their real data stop before any straight line so their conclusions is not really valid in my eyes. ( source: https://www.mdpi.com/2313-0105/7/2/28 )
View attachment 750743
Here is actual test data, 18650 NCA cells. Do not remebmer if it says Panasonic but it fore sure is cylindrical cells with NCA-chemistry.
View attachment 750744
660 days of tests(almost two years), we can see both the temperature dependent degradation as well as the SOC dependent degradation. (
3.45V = about 10-15% or so.
3.70 = about 45-50%
4.10 = about 90-95%
If I use the first clearly distinguable point on the 10C 3.7V graph and try to calculate the end point(660 days), I get 0.833% loss at day 76. using the squere root formula I end up at 2.45% degradation at 660 days and the real endpoint show about 2.2% degratation.
Of course a picture like this is har to use without fults.I did use pixel count to get as good values as possible.
For our use, the square root formula is fine. We have not the best data on our cells, as we most often have newer cells in the cars than the newest research report have. For example, the 2170L came on the market one year ago, and it probably takes one year for researches to get the hand on some of these cells and then it take another year to do a research and finalize the report.
View attachment 750748
(I also find the tremendous difference between 25C and 40C very interesting; ex: I think if I lived in a desert area like Phoenix AZ I'd be much more likely to keep the SoC around 50% during the summer than living in colder climates).
Sounds like ~ 55% SoC or below is truly ideal, but 75-90% is also definitely OK if you don't mind another ~5% degradation at 3 years, and 10% degradation at 10 years vs. baseline of 55% SoC or less. It also appears that the main difference between 70, 80, and 90% "top end charges" is really in the area of cycling aging and not calendar aging. (Other studies I've seen show 80 to 40% daily driving is easier on the battery than 90 to 50%).
Thanks again!
Similar threads
