How do the v3 charge curves make sense?

[Agon]

I keep wondering how the now famous charge curves (Tesla supercharger v2 vs v3) from the v3 supercharger make sense. It seems the SoC should increase more in the beginning when the kW numbers are larger. But actually the slope seems to be steeper between 8% and 13%. I looked at the actual data from the curves in this spreadsheet (supercharger v3 vs v2) and keep wondering how the 5 minutes between the reading at 2 minutes and 7 minutes only adds 13 percentage points (from 12% to 25%) SoC, when kW starts at 250 kW and ends at 185 kW. Whereas the 5 minutes from reading at 8 minutes and reading at 13 minutes adds 19 percentage points (30% to 49%) even though the kW numbers are much lower starting at 176 kW and finishing a little over 140 kW probably). I did a very quick and very rough calculation of the slope based on the kW numbers and a bit of guesstimation and plotted the new curve next to the actual observations (Imgur). The new curve definitely seems to suggest that the SoC should increase more in the beginning. Of course there is some margin of error in the readings and the actual kW numbers probably fluctuate a bit - and maybe he preferred to do the readings when the kW numbers were high? But I still can't make sense of this?

 

[Kant.Ing]

Maybe this is just one time result and the real life charge rates are yet to be seen or statistically analyzed or from more real-time readings.

One thing I found strange is if both under their conditions start charging from 75% SOC, they end up taking the same amount of time to reach 90%. But V3 looks to start at 70kW while V2 at 50kW there, and V3 stays always higher rate than V2. Don't add up nicely.

 

[Agon]

I am not sure you understand my argument correctly. I don't comment on the difference between the v2 and the v3, but on the difference between the curves showing kW and SoC for v3. The kW numbers dont seem to match the progress in SoC.

 

[Zoomit]

I agree that those numbers seem a little wonky; however, when I independently modeled and plotted the charging time histories, they matched up very well, surprisingly well to be honest.

Results here: Updated Model 3 Charging Profiles & Durations

 

[Shock-On-T]

I am not sure you understand my argument correctly. I don't comment on the difference between the v2 and the v3, but on the difference between the curves showing kW and SoC for v3. The kW numbers dont seem to match the progress in SoC.

You are right, the graphs don’t line up correctly.
I think we need better data.

 

[THX723]

@Agon Your first curve assumes ideal battery cells. In practice, cells and banks variance exist and take up charge differently. Charging is dynamic and cell/bank temp is closely monitored by the BMS and throttled accordingly. I seem to recall in the video the v3 charge rate bounced between 220kW and 250kW (930 and 1000 mph), despite being in the low-SOC sweet spot. It would be interesting to observe if the new "On-Route Battery Warm-up" routine would indeed optimize the charging profile closer to the ideal curve.

 

[Agon]

@Agon Your first curve assumes ideal battery cells. In practice, cells and banks variance exist and take up charge differently. Charging is dynamic and cell/bank temp is closely monitored by the BMS and throttled accordingly. I seem to recall in the video the v3 charge rate bounced between 220kW and 250kW (930 and 1000 mph), despite being in the low-SOC sweet spot. It would be interesting to observe if the new "On-Route Battery Warm-up" routine would indeed optimize the charging profile closer to the ideal curve.

I use the users actual readings of kW to calculate how much the SoC was supposed to increase based on those. So I don't assume anything reagarding battery cells. But yes, I saw some fluctuations in the kW numbers as well. And one reason might be that the user systematically wrote the larger numbers. But in the 3-4 minutes with kW numbers close to 250 kW the SoC was supposed to increase far more, as you can see when comparing the two graphs.

An example: A kW number of 250 kW is supposed to increase SoC more than 5 percentage points/min. His actual readings in the first minutes don't match that.

 

[THX723]

I use the users actual readings of kW to calculate how much the SoC was supposed to increase based on those. So I don't assume anything reagarding battery cells. But yes, I saw some fluctuations in the kW numbers as well. And one reason might be that the user systematically wrote the larger numbers. But in the 3-4 minutes with kW numbers close to 250 kW the SoC was supposed to increase far more, as you can see when comparing the two graphs.

An example: A kW number of 250 kW is supposed to increase SoC more than 5 percentage points/min. His actual readings in the first minutes don't match that.

Let’s put it another way. Not all of the stated energy (kW) is charging the cells (mph) — some are diverted for active heating or cooling. You’ll notice the stated kW and mph don’t always track perfectly (i.e. same mph at different kW). Stated kW is how much energy the car is consuming (not exclusive to battery), while stated mph is the actual charge rate of the battery. I bet a plot of MPH vs SOC would correlate better to your plot of SOC vs Time.

 

[Agon]

That makes sense and sounds like a plausible explanation for some of the difference. But as far as I remember the 250 kW was always followed by a 1000 mph number in the video clips I saw - I did not check again though. The difference has to be large to explain the difference in the curves.

 

[wwu123]

I use the users actual readings of kW to calculate how much the SoC was supposed to increase based on those. So I don't assume anything reagarding battery cells. But yes, I saw some fluctuations in the kW numbers as well. And one reason might be that the user systematically wrote the larger numbers. But in the 3-4 minutes with kW numbers close to 250 kW the SoC was supposed to increase far more, as you can see when comparing the two graphs.

An example: A kW number of 250 kW is supposed to increase SoC more than 5 percentage points/min. His actual readings in the first minutes don't match that.

You took the user's kw/kwh readings and integrated it to get time for each point, and it didn't quite match up to the user's time readings. That the integral ended up around same the final end point, 40 min for 90% SOC, rather than accumulating differences, suggests that the first and last data points are correct, but some of the intermediate values are suspect.

Putting on my best Sherlock Holmes, looking at the blank values in the tables, the computer tendency to display rounded down values (e.g. 13:59 min shows 13 min, the innate human nature to fill out spreadsheets from top to bottom, and the coffee ring stain on the bottom left, I deduce that some of the values are not actual observations but poor (generally linear) interpolations after the fact.

Basically the user filled out observations of kw vs SOC (kwh), and time vs SOC for the v3 supercharger first without gaps. then later observed a v2 supercharger, trying to match the SOC values in the first column - but the times are different, so he inserted additional rows into the time vs SOC data. That created gaps in SOC values for the v3 supercharger, which he then tried to compensate by interpolating values.

Which numbers are or are not actual observations?
-Generally a V3 SOC value in the last column that does not have the same value in the first V3 SOC column, correspond to time observations during the later V2 SOC test that didnt' have a true V3 SOC reading
-Generally V3 SOC values in the last column that have a blank V2 SOC value, correspond to time observations that didn't have a true V2 SOC reading
-V2 SOC values with a decimal point are probably also attempts to interpolate the gaps, before giving up around the 15 min mark
-But, in some V3 observations, by the time the user shifted from recording SOC to recording time, some time and SOC values incremented by 1 in the display.
-Also, the user walks with a limp due to a gruesome childhood injury, causing him to favor his weight to his left foot.

Elementary, my dear Watson...

 

[acoste]

I use the users actual readings of kW to calculate how much the SoC was supposed to increase based on those. So I don't assume anything reagarding battery cells. But yes, I saw some fluctuations in the kW numbers as well. And one reason might be that the user systematically wrote the larger numbers. But in the 3-4 minutes with kW numbers close to 250 kW the SoC was supposed to increase far more, as you can see when comparing the two graphs.

An example: A kW number of 250 kW is supposed to increase SoC more than 5 percentage points/min. His actual readings in the first minutes don't match that.

Agon, you are right. The SoC vs time doesn't make sense. I calculated the derivative of it, see below. Even the V2 is wrong. Look at V2 peaks and the flatness above 40 minutes.

What is the source of this data?

 

[Agon]

The source is a Reddit user who attended the v3 event. I have no intention of casting doubt on his conclusions, but when several serious media suddenly cite those numbers as evidence of how a very important new technology works for a large listed company, I will try to make sure that those numbers are solid.