[Poll + Tracking] HV battery state of health tracking (SOH) w/ self updating analytics -- crowdsourced data (add yours)

[Mache_Nor]

Forgot to mention; Thanks for having the number of reports listed! I totally forgot about promising to help you out (sorry), but seems you managed to find a way meanwhile

Sponsored

 

[flapjake314]

it has been about 3-4 months since i started this thread and many of the datapoints were entered, so i'll use this as an opportunity to remind folks to collect/enter a new datapoint to track themselves over time, especially now that we're in the winter months

 

[flapjake314]

my personal trendline after i've started tracking as part of this dataset. heading straight toward 70% SOH at the end of the warranty period!

google: how do i accelerate SOH deterioration so i get a warranty repair

 

[Mache_Nor]

You did the cold weather reading now right? Let's see how it turns out next summer. If equally bad, then it's fast charging 1-100% every day, keep it at 1% most of the time, and make sure to never precondition or do anything nice to it (even say offensive stuff to your pony so that it's feelings might get hurt, which should result in it a worse degradation of its batteries) Then it's yolo until hopefully 69% (nice) before the warranty runs out.

 

[flapjake314]

You did the cold weather reading now right? Let's see how it turns out next summer. If equally bad, then it's fast charging 1-100% every day, keep it at 1% most of the time, and make sure to never precondition or do anything nice to it (even say offensive stuff to your pony so that it's feelings might get hurt, which should result in it a worse degradation of its batteries) Then it's yolo until hopefully 69% (nice) before the warranty runs out.

yeah i did the reading recently at around 32F / 0C. the [unfortunate] thing is there is actually a strong trendline in the readings since i first took them this summer when it was warm lol. i'm sure there it will level out over time etc, but right now there is a strong pattern

i have no unique driving situations other than i'm relatively low mileage (~7000mi/year). even then it's not incredibly low, if folks think the "normal" usage is more like 10k

edit: you know what i'm realizing would be a very interesting graph? instead of looking at the trend of the whole population, to look at the trends individually. would need lots of data from the same individuals over a longer period of time, but something i'll keep in mind...

 

[Mache_Nor]

yeah i did the reading recently at around 32F / 0C. the [unfortunate] thing is there is actually a strong trendline in the readings since i first took them this summer when it was warm lol. i'm sure there it will level out over time etc, but right now there is a strong pattern

i have no unique driving situations other than i'm relatively low mileage (~7000mi/year). even then it's not incredibly low, if folks think the "normal" usage is more like 10k

edit: you know what i'm realizing would be a very interesting graph? instead of looking at the trend of the whole population, to look at the trends individually. would need lots of data from the same individuals over a longer period of time, but something i'll keep in mind...

That actually is very interesting! I will do a yearly reading, but we will have to have some sort of structure here to sum up the individual readings, then collect / group them together. That way we can be able to predict better. But how … uff, too much food this Xmas for functional brain.

Edit: since all answers are collected in same form, should it not be possible to select each answer and filter on that? Like the normal sorting / filter function on columns in Excel? Since the list have all we sort of already have the collective results, now it’s just to be able to see each trend so we can do insights on them. Alternatively, each answers can be stored in each page / tab / table, so that repeatable answers contributes to your individual readings.

 

[Trick.Mach-E]

Our HVB had jumped to 93.5% at some point after my last post and today our HVB is showing 94%.

Poll updated...

<EDIT 10/15/24>

Battery SOH is now at 93% at 45640. Recently my wife drove 150 miles a day for a few days in a row. She began each day with the battery at 100% drove the roughly 150 mile round trip returning home in the 40% range... each night a full charge to 100% for the next day...

<Original Post 10/07/24>


I finally added my data to the poll. Our FE has 45074 miles (battery age is 47 months old) and the SOH is at 92%. It had fluctuated a little bit between 91% and 92.5% over the past couple of thousand miles. The 91% caught my attention just as much as the 92.5%...

I really haven't done much to keep the battery in optimal condition with long term reliability in mind... We DCFC a lot on the car as well as we take it on 320 mile trips to Los Angeles and back (160 miles each way) and these trips often has us going from a 20-30% charge to 80% on the way back home.

Around town I plug in when it gets low (no set number I just wing it) and then charge to 80% with our ChargePoint Home Flex.

 

[peichd1]

if anyone has ability in statistics, i'd like to understand how the 4 variables are correlated (and how strong the correlations are) within the dataset: SOH, mileage, age (currently proxied by model year), and DCFC frequency.

i of course know how to do it with 2, i think know how to show it with 3 (that's my bubble graph), i have no idea what to do with 4. and based on that maybe we then think about which questions we add / remove etc

the other thing i haven't figured out how to make useful is some kind of standard deviation metric, to another poster's point about knowing where on the range you sit. i have the trend line and of course can calc the standard deviation but i don't know how to make it easily useful for any forum member

I have access to a multivariable regression package, so I could do some statistical analysis on the relative significance of the different input variables. Where can I access the raw data?

 

[flapjake314]

I have access to a multivariable regression package, so I could do some statistical analysis on the relative significance of the different input variables. Where can I access the raw data?

it's just in tab 2 of the google link. first tab is the graphs but 2nd tab is all the data.

 

[peichd1]

I ran a multivariable regression analysis with the data collected so far and found that just fitting to the log of the battery cycles provided the best fit of the data. Adding the other variations did not provide much improvement of the fit. Fitting to the battery age was also good, but not as good at the battery cycles.

The fit is not great, there appears to be a fair amount of "noise" in the data. If I subtract the reported SOH from the fitted equation, the error has a standard deviation of about 2%. The error could be from not knowing the true age of the battery pack, error in the SOH routine, or maybe temp effects on the SOH estimation. I've seen my SOH drop 6% during the cold weather in Minnesota, hoping I see that bounce back in the summer.

Not sure it makes sense to me the fit vs log battery cycles makes sense, it would suggest that the aging slows down vs time which I don't think is what we expect. Seems like there is an initial drop, is that just some sort of battery breaking in thing or is it due to the delay from battery manufacture to when you actually get the car? Could the log fit be more due to people with higher usage have more battery cycles but not more calendar gaining, so the data makes it look like the battery aging slows down. Seems like that would have came out in the regression analysis as the battery age providing a bigger improvement to the fit.

 

[flapjake314]

I ran a multivariable regression analysis with the data collected so far and found that just fitting to the log of the battery cycles provided the best fit of the data.

i had a DM chat with peichd1 already before he presented the results, public thanks for his work. my take-aways were:

1. cycles explains the results better than battery age; age didn't add much to the trend. so low mileage drivers should still have more SOH than higher mileage drivers of the same age

2. cycles is obviously more important than miles. we use miles because it's easier to see and measure, but the more efficiently you drive (and the larger your battery) the better your SOH will be at the same # of miles travelled

 

[Elar]

Hello
I would like to ask for advice.
My almost 3 year old car (40.000km) has shown 95% HvBSoh so far. This is good.
It has now been in the service for 2 weeks. The car has been parked for almost 2 weeks (akku at 40%), between -7 and +5 degrees)
They have changed the HVJB. They charged it up to 100% and got it back.
I checked today and the SOH is 86%...
This shocked me.
What is the explanation for this? Why dropped 9%????
The other values that are showing (currently) on this car:
HvBSoc: 61.31% (car display shows 66%)
Energy to empty: 46.91kwh

AWD ER (2022)

 

[flapjake314]

Hello
I would like to ask for advice.
My almost 3 year old car (40.000km) has shown 95% HvBSoh so far. This is good.
It has now been in the service for 2 weeks. The car has been parked for almost 2 weeks (akku at 40%), between -7 and +5 degrees)
They have changed the HVJB. They charged it up to 100% and got it back.
I checked today and the SOC is 86%...
This shocked me.
What is the explanation for this? Why dropped 9%????
The other values that are showing (currently) on this car:
HvBSoc: 61.31% (car display shows 66%)
Energy to empty: 46.91kwh

AWD ER (2022)

Your post is confusing. You start by stating 95% SOH and end by saying 61% SOC (or 85% SOC). are you asking about SOH or SOC? Which number is which?

 

[Elar]

Your post is confusing. You start by stating 95% SOH and end by saying 61% SOC (or 85% SOC). are you asking about SOH or SOC? Which number is which?

Sorry.
Starting 95% SOH - HVJB - 86% SOH

 

[mkhuffman]

Sorry.
Starting 95% SOH - HVJB - 86% SOH

Such a large drop is unusual. I recommend you try recalibrating the BMS by following the procedure outlined by @Mach-Lee. You should search the forum for his post but essentially it involves running the battery down to around 10% (lower is OK if you want) and charging it back up to 100%, and then running it back down to 10% again.

By the way, when you charge it to 100%, check the EtE and see what the value is. If you calculate the SoH using EtE (I assume you have the 91-kWh battery pack) you should get something close to the SoH of the battery. I have found SoH is usually a little worse than the EtE I can actually charge the battery to at 100%, but it correlates well.

Sponsored