Battery health tracking SS

[Chrushev]

One thing I dont understand.

At 100% charge it shows 298 miles which is ~5.5% degradation (using 316 miles even though mine was always 310 as new).

When we look at the distance mode data, if I use the 77.8 kWh figure, then I have 8.6% degradation, but if I use the 75 kWh figure then it lines up almost perfectly with the car's 298 mile... ~5.2% degradation. (75.4kWh would make them line up perfectly)

So one of these is not correct. Which one?

 

[AlanSubie4Life]

One thing I dont understand.

At 100% charge it shows 298 miles which is ~5.5% degradation (using 316 miles even though mine was always 310 as new).

When we look at the distance mode data, if I use the 77.8 kWh figure, then I have 8.6% degradation, but if I use the 75 kWh figure then it lines up perfectly with the car's 298 mile... ~5.2% degradation.

So one of these is not correct. Which one?

Degradation threshold. If you post the picture requested we can show how that works, as mentioned above. Does not have to be at 100% (just more accurate). Three numbers, one with % as before, the other with rated miles/km.

And yeah it is 316 not 315.

Initial capacity loss of a couple percent (approx, depends on vehicle) is hidden from view. You don’t see max miles reduce until you get below the degradation threshold (there is not a top buffer exactly; the miles expand in energy content, and immediately start getting used when driving). I don’t view this as a buffer though it is in a sense of course.

The approximate 73kWh (need to see the energy screen to know what the number actually is - 73kWh is extrapolated from other data, making assumptions (namely that degradation threshold is 77.8kWh which it probably is not), and may actually be incorrect), divided by 78.5kWh is correct. Within the errors inherent in the system.

If degradation threshold is e.g. 76kWh then 71.1kWh would be roughly 76kWh*298/316. But you still start at 78.5kWh.

So you might be at 71kWh not 73kWh. Need to see that vehicle constant or the energy screen at close to 100%.

I found an old post showing rated line for 2020 Model Y LR to be at 246Wh/mi so the constant should be 241Wh/mi and degradation threshold about 76.2kWh (241Wh/mi*316mi).

That makes most of the numbers align fairly well but more pictures are welcome.


Actual rated wh/mi

I’ve learned a couple things since then so what is there is not exactly correct but it is close.

So summarizing:
FPWN, full pack when new: 77.8kWh
Actual full pack: ~78.5kWh
Rated constant 2020 Model Y LR: 241Wh/mi (energy screen line is at 246Wh/mi, which is expected, always 5Wh/mi high)
EPA range: 316 miles
Degradation Threshold: 316mi*241Wh/mi = 76.15kWh
Usable when new 95.5% of 78.5kWh = 75kWh

Your car specifics: 248Wh/mi * 215mi at 75% => 71.1kWh (some rounding error)
298 rated miles at 100% (better data point) => 298/316*76.15kWh = 71.8kWh

Capacity loss: 71.8kWh/78.5kWh = 8.5%

 

[AAKEE]

This is for 21 inch wheels Performance model.

@AlanSubie4Life is right.

At that time there was only one pack for the LR/P: 77.8kWh marked size.
It has 4416 pieces of Panasonic 2170 cells.
They often delivered more energy, just like @alansubie said and showed.

The wheel size does’nt matter. In theory a larger wheel will induce more need for power, which increases the losses. For the low power needed in the EPA test the difference is probably neglible.

 

[AAKEE]

I wonder if my capacity was on low end

There is/has been a lot of talk of the battery lottery.

From my perspective there is very minor differences between different packs.
We se slight variations in the EPA tests but they most often is well within 1%. The car stands fully charged overnight before the EPA test and different imbalances between the tests probably causes small difference in energy burned to balance overnight.

The close numbers between the tests tell is that the packs start at similar capacity.

Your pack has degraded just as expected.
Your 71kWh is slightly above my calc, but on the other hand I did set the origin at 77.8 (specified size in the BMS) instead of the measured around 78.5kWh.
If I adjust to 78.5 the calc will be within 1% from your value.

Eben if it is hard to know the avg. cell temp the best effort made to get your numbers shows to me that your car has degraded just like expected in your cars environment and usage.
If you had ten Y LR that ypu used the same the capacity and degradation would be within 1% from yours.

What we can see in your case is very good.
Everything is as expected.

We never got to the discussion about How degradation progresses but calendar aging takes the main bite and it reduces the rate quite fast. You could count on to be ”around” 10% degradation for long time.
So it is time to be happy now

 

[Chrushev]

Here is consumption screen at 98%

These numbers indicate 71.386 kWh?

Also, you said "degradation threshold" several times. What is that?

 

[Chrushev]

There is/has been a lot of talk of the battery lottery.

From my perspective there is very minor differences between different packs.
We se slight variations in the EPA tests but they most often is well within 1%. The car stands fully charged overnight before the EPA test and different imbalances between the tests probably causes small difference in energy burned to balance overnight.

The close numbers between the tests tell is that the packs start at similar capacity.

Your pack has degraded just as expected.
Your 71kWh is slightly above my calc, but on the other hand I did set the origin at 77.8 (specified size in the BMS) instead of the measured around 78.5kWh.
If I adjust to 78.5 the calc will be within 1% from your value.

Eben if it is hard to know the avg. cell temp the best effort made to get your numbers shows to me that your car has degraded just like expected in your cars environment and usage.
If you had ten Y LR that ypu used the same the capacity and degradation would be within 1% from yours.

What we can see in your case is very good.
Everything is as expected.

We never got to the discussion about How degradation progresses but calendar aging takes the main bite and it reduces the rate quite fast. You could count on to be ”around” 10% degradation for long time.
So it is time to be happy now

The graphs you posted before (unless I missed one) all has battery cycles on X axis. is there one for Age on the X axis?

 

[Chrushev]

Also, have you tried the Tesla's battery test? How accurate is it?

 

[AAKEE]

The graphs you posted before (unless I missed one) all has battery cycles on X axis. is there one for Age on the X axis?

No, there is one chart for Calendar aging as this is tested separately and one for cyclic aging.
You can not easily show time, temperature, SOC, cycles number and maybe depth of discharge in one two dimensional chart.

Try to look little in the link to the resesrch report I posted. It is very thorough but just take a look ans see how the tests are made and how the result is presented.

 

[Chrushev]

No, there is one chart for Calendar aging as this is tested separately and one for cyclic aging.
You can not easily show time, temperature, SOC, cycles number and maybe depth of discharge in one two dimensional chart.

Try to look little in the link to the resesrch report I posted. It is very thorough but just take a look ans see how the tests are made and how the result is presented.

How do you plot out your expected degradation ? You posted a screenshot of your spreadsheet but it doesnt show what is on the axis. Are you using cycles for the X axis there too? The problem with cycles on X axis is that for some people 200 cycles is 1 year, for others 50 cycles is 1 year.

 

[AAKEE]

How do you plot out your expected degradation ? You posted a screenshot of your spreadsheet but it doesnt show what is on the axis. Are you using cycles for the X axis there too? The problem with cycles on X axis is that for some people 200 cycles is 1 year, for others 50 cycles is 1 year.

I will not reveal the complete calculation.

Calendar aging is calculated from ”average SOC” over a normal day. It includes for example data points from the time of the day where you normally start charging and the estimated average temperature of the pack. (Average ambient plus some degrees for charging/driving. I have > 3 years logs of ambient temp + average cell temp).
End of day SOC is input, which can be calculated from the input data.

Cyclic aging is depth of discharge (charge level - end of day doc) and where in the SOC regime it happens.
Number if cycles is calculated with in-data, I normally ask for lifetime average consumption. The new battery capacity/lifetime consumption gives a 100-0% range so the Full Cycle Equivalent (FCE) of for example 21K miles can be calculated.
It will be the total ODO number that set the cyclic degradation, how much each year is not important here.

Charts for cyclic aging is normally created with FCE on the X-axis.

The degradation per cycle is made by grabbing number in a table I created.

There is a few twists further buth thats the big picture.

 

[AAKEE]

How do you plot out your expected degradation ? You posted a screenshot of your spreadsheet but it doesnt show what is on the axis. Are you using cycles for the X axis there too? The problem with cycles on X axis is that for some people 200 cycles is 1 year, for others 50 cycles is 1 year.

Just realized you mean my chart for my own cars expected degradation?

Thats calendar aging only. Using low SOC cycles the cyclic aging will be so small that it basically can be disregarded.
At 55-45% as most days it ”costs” about 5% for 1000 FCE = thats around 485K km (300 K mi) with my car with my average.
So, the cyclic aging will be around 0.2% each year, driving 20K km.

 

[AlanSubie4Life]

Here is consumption screen at 98%

These numbers indicate 71.386 kWh?

Yes. Would be good to get a similar screen showing miles so we can calculate the constant. While I have that prior data point of 241Wh/mi, I can’t really make it fit your captures, exactly.

Close (within 1kWh), but something’s not quite right.

So a picture like this then flip the mode and one more. Taken in quick succession.

Also, you said "degradation threshold" several times. What is that?

It’s the full pack energy level below which your vehicle starts to show loss of range. It’s between ~76kWh and ~77kWh for your car.

Also, have you tried the Tesla's battery test? How accurate is it?

It’s probably fine, but unfortunately the denominator seems to be a bit arbitrary. This tends to make it optimistic.

Anyway there is no need to run it. Answer is 90-91% health. It will probably say 93-94%.

 

[DarrenHD]

I have noticed since 2024.14.x (and up until 2024.20.x) my range estimates using TeslaMate and TeslaFi have dropped about 5-6%. I wonder if this is due to changes Tesla has made to the BMU and battery health estimation. I've lost about 20-30km of range since 2024.14.x with really no change in my charging or driving habits.

 

[AAKEE]

Also, have you tried the Tesla's battery test? How accurate is it?

View attachment 1060785

This is the thing:

The slower you discharge the more energy you get out of the battery. This is not a new fact and has been known since the first battery.

The EPA test is relatively slow discharge, but perhaps equals driving 55mph or so.

The built in battery test that we *By coincidence* can access has a very low discharge power (using the heater/heat pump etc).
The result is higher than the estimate the BMS does.

The BMS estimate is very close to my consumption when driving gentle highway driving. I have compared the nomimal energy on board before and after the drive with the consumed energy according to the trip screen and more or less always have a very good match. This, for both my former M3P and the MSP.

So, as it seems the capacity estimate of the BMS equals about the EPA test discharge rate. And my consumption of the battery, so for example using 50kWh equals that portion of the battery pack energy.

To make it worse, the energy test finds a lower energy delivered to be 0% degradation.

This means, Tesla use a lower number than the real starting capacity (compared to what the car actually delivered in the EPA test) as the zero degradation mark.
And they discharge slowly so the battery can deliver more energy.

I can supply with several examples of these tests made, with a lower than usual degradation.
After checking with the energy graph the BMS estimate is a couple of kWh lower than the result from the battery test.

Of course Tesla knows that. Planting a hidden test that shows lower degradation will make a lot of people happy (and reduce the number of questions about degradation to Tesla).

 

[AlanSubie4Life]

I have noticed since 2024.14.x (and up until 2024.20.x) my range estimates using TeslaMate and TeslaFi have dropped about 5-6%. I wonder if this is due to changes Tesla has made to the BMU and battery health estimation. I've lost about 20-30km of range since 2024.14.x with really no change in my charging or driving habits.

Could be coincidence. It’s not unusual for the estimate to adjust in a large step.

Could also be the update, but the vehicle itself will still tell you the health and most likely be correct, so you can always check that.

 

[AAKEE]

This is the thing:



I can supply with several examples of these tests made, with a lower than usual degradation.
After checking with the energy graph the BMS estimate is a couple of kWh lower than the result from the battery test.

This is one example, so far I have seen at least 20-30 all showing the same thing;

The battery test: 3% loss

The BMS value (extracted by reverse maths but we know a look into the BMS would say the same) and we know the BMS value very well reflects the energy we get out of the battery by driving the car.

74.5kWh, out of 78.8kWh ”spec” which often reach above 79 kWh.

Thats at least 5.5% real degradation.

 

[AlanSubie4Life]

Could also be the update, but the vehicle itself will still tell you the health and most likely be correct, so you can always check that.

To be clear: I don’t mean the health test here, I just mean reading the capacity from the energy screen.

 

[Chrushev]

This is the thing:

The slower you discharge the more energy you get out of the battery. This is not a new fact and has been known since the first battery.

The EPA test is relatively slow discharge, but perhaps equals driving 55mph or so.

The built in battery test that we *By coincidence* can access has a very low discharge power (using the heater/heat pump etc).
The result is higher than the estimate the BMS does.

The BMS estimate is very close to my consumption when driving gentle highway driving. I have compared the nomimal energy on board before and after the drive with the consumed energy according to the trip screen and more or less always have a very good match. This, for both my former M3P and the MSP.

So, as it seems the capacity estimate of the BMS equals about the EPA test discharge rate. And my consumption of the battery, so for example using 50kWh equals that portion of the battery pack energy.

To make it worse, the energy test finds a lower energy delivered to be 0% degradation.

This means, Tesla use a lower number than the real starting capacity (compared to what the car actually delivered in the EPA test) as the zero degradation mark.
And they discharge slowly so the battery can deliver more energy.

I can supply with several examples of these tests made, with a lower than usual degradation.
After checking with the energy graph the BMS estimate is a couple of kWh lower than the result from the battery test.

Of course Tesla knows that. Planting a hidden test that shows lower degradation will make a lot of people happy (and reduce the number of questions about degradation to Tesla).

Tesla’s test says to run with less than 50% SoC. And says test takes 24 hours to run.

So does it discharge the battery fully then charges it up all the way? If you run the test at let’s say 2% SoC how can it judge heath from discharging 2%?

At the same time the car should know how much it pulled to total for a 100% charge and it should also know the overhead wasted, simple subtraction math should give it pretty accurate result that would not depend on discharge speed.

 

[AlanSubie4Life]

Tesla’s test says to run with less than 50% SoC. And says test takes 24 hours to run.

So does it discharge the battery fully then charges it up all the way? If you run the test at let’s say 2% SoC how can it judge heath from discharging 2%?

At the same time the car should know how much it pulled to total for a 100% charge and it should also know the overhead wasted, simple subtraction math should give it pretty accurate result that would not depend on discharge speed.

It discharges it. But actually I have no idea. In any case it is a waste of time and a battery cycle. You already have your answer, and we also know health test result.

 

[AAKEE]

Tesla’s test says to run with less than 50% SoC. And says test takes 24 hours to run.

So does it discharge the battery fully then charges it up all the way? If you run the test at let’s say 2% SoC how can it judge heath from discharging 2%?

At the same time the car should know how much it pulled to total for a 100% charge and it should also know the overhead wasted, simple subtraction math should give it pretty accurate result that would not depend on discharge speed.

EPA-test, snd also the WLTP-test.

-charge full
-let car stand overnight
-drive until it stops
-recharge and measure energy from net to fully charge.