Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Displayed Range and Seasonality

This site may earn commission on affiliate links.

FlasherZ

Sig Model S + Sig Model X + Model 3 Resv
Jun 21, 2012
7,030
1,032
I posted some rough data in the 90% thread and the 90D thread, but felt it might be appropriate to pull another thread here when discussing range and degradation.

It's starting to get colder here in the midwest, and my displayed range after charge (at 90% and 100%) has been dropping again, as it has for the past few years when the colder weather rolls in. I've also noticed it picks back up in the spring.

So I put my database of information from my car to good use (it dates back to October '13, data corruption destroyed the DB prior to that date) and graphed my range for both 90% and 100% charge levels.

Here are the two graphs. First, 90% (selection criteria - all points, API battery_level = 90):
charge90.png


Next, 100% (selection criteria - all points, API battery_level = 100 (that's right, I am not afraid to range charge!):
charge100.png


As I noted in the other threads, I have not done any correlation with temperature, distance from range charge / rebalance, or software updates. I did go back and look at the anomaly that occurred on late Feb, 2014, when the graph does a correction upward. After thinking about it - this was the date that my rev A battery pack was replaced with the refurbished rev D pack following the contactor failure.

Note that this is not the same thing as the snowflake that "reserves" some range in ultra-cold temps - this is the car's interpretation of the range of the battery.

Conclusions to reach here: range display is seasonal and as you enter winter months, you may not be experiencing the degradation that you think is appearing... it's likely that it comes back in spring.
 
Any chance you can overlay the 10/13-10/14 data with the 10/14-10/15 data to see the YoY and MoM changes?

I did a couple of overlays but the points are a bit too noisy, I tried a few trendlines (6-order polynomial, moving average, etc.) and nothing produced anything that appeared to be worth anything. I suspect you're going to need weather as a correction factor to do any sort of overlays.

Here's what a moving-average overlay looks like. Blue is 2/25/2014-2/24/2014, Green is 2/25/2015 onward:

chargeoverlay.png
 
Another view... here is the monthly average range at 90% charge (mean of all values within a month beginning with the date on the X axis). EDIT: I completely forgot I had an easy source of comparison with regard to temperature. I have the car's "outside_temp" data. Here's an overlay of the average outside temperature (as registered by the car, deg C, unheated garage) with the average range:

range-v-temp.png
 
Last edited:
I did a couple of overlays but the points are a bit too noisy, I tried a few trendlines (6-order polynomial, moving average, etc.) and nothing produced anything that appeared to be worth anything. I suspect you're going to need weather as a correction factor to do any sort of overlays.

Here's what a moving-average overlay looks like. Blue is 2/25/2014-2/24/2014, Green is 2/25/2015 onward:

You're right, it is too noisy to do any real comparison. But it does have a nice trending pattern and there are areas of the graph that do overlay with each other.


Also, this doesn't take into account any potential changes to the algorithm that Tesla may have done with a firmware push.
 
One more - the full data series for range vs. temp, rather than average. There are roughly 55,000 temp samples and 11,000 90% range samples over the time period:

range vs temp.png


(You can also see where the holes in my data are due to the sampling getting "stuck".)
 
Last edited:
So I guess my question is why. Why does rated range correlate to ambient temp? Sure, when it gets colder less energy is available to draw out, but the snowflake should illuminate in this instance. Furthermore, if these reading are taken immediately on charge completion (as they are API readings), the pack should be fairly warm and ambient temp should not hold influence.
 
So I guess my question is why. Why does rated range correlate to ambient temp? Sure, when it gets colder less energy is available to draw out, but the snowflake should illuminate in this instance. Furthermore, if these reading are taken immediately on charge completion (as they are API readings), the pack should be fairly warm and ambient temp should not hold influence.

I'm not sure, I'm not a lithium battery or BMS expert by any means.

The snowflake is used to represent Tesla's concern that rapidly-dropping temperatures could leave the battery below cut-off voltage. I've only seen it on my car once or twice since they introduced it. I've always understood the snowflake to be a "reserve" held in case the pack gets much colder (e.g., a big difference between pack temperature and external ambient temp). I don't have enough experience with it, though.

- - - Updated - - -

FlasherZ:
Just an observation, there are peaks in the summer and dips in the winter.... my goodness.... this just looks exactly like the Miles Per gallon chart that my Prius IV shows on Fuelly.com

Good Work.

The range peaks are actually occurring in the spring, and you see a reduction in the summer months here (our hottest months tend to be Jul-Aug, although this year's data is an anomaly because we spent 2 weeks vacationing in Florida - remember these are car temps).
 
So I guess my question is why. Why does rated range correlate to ambient temp? Sure, when it gets colder less energy is available to draw out, but the snowflake should illuminate in this instance. Furthermore, if these reading are taken immediately on charge completion (as they are API readings), the pack should be fairly warm and ambient temp should not hold influence.
Temperature definitely influences rated rage. at 100 degree you get more out than at 50 degree. But the car won't show it as a difference on the battery meter. That only seems to happen when it gets really cold. But the car does know it can get more out of the batter at warmer temperatures and shows it as rated miles.

Charging is pretty efficient. There isn't too much wasted heat. But either way, no matter what the ambient temperature is, charging will bring up the battery temperature a little. But it does so when it's 75 as much as it does when it's 45.
 
Temperature definitely influences rated rage. at 100 degree you get more out than at 50 degree. But the car won't show it as a difference on the battery meter. That only seems to happen when it gets really cold. But the car does know it can get more out of the batter at warmer temperatures and shows it as rated miles.

Charging is pretty efficient. There isn't too much wasted heat. But either way, no matter what the ambient temperature is, charging will bring up the battery temperature a little. But it does so when it's 75 as much as it does when it's 45.

The data I've posted today does suggest that temperature plays a role in the rated range seen after charging (a/k/a the "battery meter"). It doesn't have an effect on the percentage, of course -- if I graphed that, it would be a straight line at 90% and 100%, because these samples only come when the car thinks the pack is at 90% or 100%. However, the car's "rated range" calculated from that is obviously differing. Right now my maximum tends to be 234ish for 90%, whereas in the summer it tends to be 238ish (and you see some data points that go up to nearly 241) and it is cyclic. Each winter I wondered if I was degrading, and yet each spring I would recover it.

For those who didn't understand it in my posts above: I'm taking all the "rated range" samples as reported by the API ("battery_range") when the car reports that it is at 90% SOC (top graph) or 100% SOC (bottom graph) - generally within 15 minutes after charging (the car goes to sleep after that and TeslaMS stops sampling). So this is the "rated range" displayed by the car at 90% SOC.

The temperature portion of the graph is a collection of the samples all taken while the car is awake. Unlike the charging data, it is a collection of ALL temperature samples, not just the ones at 90%/100% SOC.

EDIT: Removed correlation coefficient - calculation errors.
 
Last edited:
Charging is pretty efficient. There isn't too much wasted heat. But either way, no matter what the ambient temperature is, charging will bring up the battery temperature a little. But it does so when it's 75 as much as it does when it's 45.

Agree that the heat from charging will be the same regardless of ambient temp. But the BMS will trigger pack heating at ~8 C = 46 F so the argument that the pack is much colder does not stand.

Separately, I can tell you that my rated range hasn't budged from when we had nighttime lows in the mid-70s two months ago and the lows of mid-40s these days. So a much greater difference must be needed.
 
This is great, thanks Flasher for sharing your data with us!

I have a lot of data on my 90% and 100% charges down to 2 decimals from January till now that I can share as well, but need a bit of time to format it properly to be significant like yours.

In summary -- and I know it's been less than a year, but -- I don't personally see seasonality in my rated range readings. They've mostly just gone down more or less steadily (and recently, much faster unfortunately).

Quick question: are all your 90% and 100% readings done with RANGE MODE in the same position (either ON or OFF)? I ask because turning on range mode adds about 3 miles of rated range, at least for my model (P85D).
 
Quick question: are all your 90% and 100% readings done with RANGE MODE in the same position (either ON or OFF)? I ask because turning on range mode adds about 3 miles of rated range, at least for my model (P85D).

Yes - 'range mode' is turned off, and power savings is on.

We tend to have broader range of weather here (everything from -5 degF to 105 degF typically), so I would expect those who live in more stable weather environments to see this effect less if the hypothesis holds. I would love to see others' data along these same lines.
 
Sure, but the battery, at least while charging, doesn't see -5. It sees more like 45 which is the active heating target and this would be the same regardless whether it is 45 F or -10 F. What am I missing?

I don't know, really. I just found it interesting that in the winter the car will calculate that its rated mileage is lower at the same SOC. Perhaps that formula includes ambient temperature, perhaps the temperature is sensed in the charger instead of the battery, it could even be temperature effects on electronic components... More questions than answers, I'm afraid.
 
In a quest to find tighter correlation, I did some more work. This time, I pulled temperature samples only within 30 seconds of the 100% SOC range samples I had pulled.

range vs temp 30sec.png


The correlation is now tight, with only two explainable areas of disconnect: 3/2014, just after the new pack had been installed; and 8/2015, just after my car's energy data had been reset due to the accident repairs from colliding with a deer.