@Zoomit, have you ever tried charting the profile with the x-axis labeled as % still, but using a time-scale?
If so, could you point me to one? It occurred to me this may perhaps be another interesting way to look at the charge profile.
Hmm, or will it be interesting? I can't picture it ... I think it would be? LOL.
If you don't have timestamps, you could approximate time based on the average power between any 2 data points and call the time delta, dt = [dX (%) * Capacity (kWh) / P (kW)] at any given data point? e.g. dt = dX/P. P = dX/dt ... are we doing calculus now? Would this actually be adding any data/info to the visualization? Hmmm... or would total kWh added vs time be more interesting? Hmmm. No, I think stretching out the x-axis labelled as % to the time scale would be informative for the viewer. Not as a replacement to the existing charts, but as a supplement.
On a side-note, I wonder how much of the individual session charge profiles depends on time, and not just SoC, ignoring other variables like temperature. Like even if temps are identical, after X minutes has passed at max power (or whatever power rate) does the car decide to taper regardless of all other factors being the same? The convergence after certain percentages like 60%-ish lead me to believe that's not the case at least for that range above 60%, but perhaps this type if time-based, temperature-based, SoC-based combined logic will never lead to individual profiles matching each other closely unless everyone starts at exactly the same SoC? I guess this is where ABRPs large number of datapoints and charting the medians helps paint a better 'average' (or median) picture.
Here's a quick and dirty (heh, no pun intended) example using
@DirtyT3sla's "data and ugly chart" ... top chart is the original [modified to end at 90% to match 2nd chart], bottom is with my modified x-axis using time* (as calculated by method in prior post) and then labelling the x-axis using the SoC column.
I think this would be more interesting on charts that show the ramp, higher peak and longer high-rate section and then a taper.
View attachment 421502
*Formulas used for time:
dX (%) = SoC minus prior_SoC
dE (kWh) = dX * 75 kWh
dT (h) = dE * 2 / (power + prior_power) [e.g. Energy / AveragePower]
Yeah, essentially. But instead of SoC as a separate line graph, the x-axis time labels are %SoC instead of time. Same curve shape as your first chart. Comparing those two, the SoC-scale taper looks linear and time-scale curved. I think that is some extra visualization that’s useful.
It may be difficult to overlap multiple charts in a way to meaningfully compare them though. That would probably be best (on a time scale) with your SoC line? Or maybe the time-scale is just for standalone charts.
The current charts are probably a great way to compare multiple charts to try to flesh out a profile, assuming it’s mainly dependent on SoC, and not on current session duration (assuming other factors like preconditioning and temperature are constant).
Ok, lot's to unpack here
@darth_vad3r. The charge profiles are not a direct function of time. For our purposes here, they are primarily a function of voltage, amperage, battery temperature and Supercharger temperature. The battery temperature is a function of internal resistance, initial temperature, cooling system performance and time. In addition, the Supercharger temperature is a function a bunch of similar parameters including starting temperature, cooling system performance and time.
But to
best capture the charging profile I use battery level (%), more accurately called percent indicated state of charge,
and power (kW). These are rough surrogates for the voltage and amperage parameters needed to define the charging profile. We don't have any quantitative temperature data but look to ambient temperatures and ORBW use to gauge battery and SC temps. As expected, the V2 charge rates are heavily dependent on these temperatures.
But "time" is critical to interpreting charging performance. The primary two metrics that users care about are range and time. How far can they drive given a number of minutes of charging. Power, energy, temperature are just geek details.
To date, I've collected 30-40 charging sessions from various cars to calibrate my charging models. Integrating the charging power to get energy vs time would be arduous for all these sessions. My interest is in determining a reference profile, aka ideal profile, for a particular car/configuration to be able to compare that to other cars/configurations. It's creating and using a common yardstick that is valuable for these comparisons, not comparing particular charge sessions with random performance. The ABRP data shows how much variability there is to Tesla charging. My goal is to get away from that randomness by focusing on the ideal performance of the different cars.
So I have created ideal charging profiles and, as you suggested, integrated the power to get energy over time and plotted them all different ways. The ones that are most useful I've attached below. For comparison's sake, I'll use a Model 3 LR and an e-tron, since it's getting a fair amount of attention lately. One important note about the Model 3 profiles--these are a work in progress, hence this and a
previous thread trying to collect and understand real-world data.
Here are the charging profiles shown similarly to previous graphs in this thread: kW vs %SoC.
Here they are with kW vs time. This is noteworthy because the area under each curve is equal to the usable energy of the battery. However, this graph over-emphasizes the lower power portion of each profile and makes details at high power harder to discern. An important point for the "250kW" charging Model 3 is visible in this chart: 250kW is only seen for 4 minutes...and that's when starting at 0%.
So now if we integrate power over time we can see the cumulative energy gain over the duration of a charge session from 0% to 100%. As seen on the previous graphs, the e-tron battery charges similarly to the the Model 3 on a 150kW V2 Supercharger. The dashed blue and green lines are similar for the first 2/3rd of the charge session.
Ok, thanks for staying with me here. This is where it gets relevant to people like my wife, who don't give a lick about kW's or kWh's. The normal driver needs to know range vs time. But before we get there, we need to convert the energy above into a vehicle specific EPA range. Instead of a Model 3 LR
battery, I'm making reference to Model 3 LR
AWD/P EPA data below. This first graph has charge rate in terms of mi/hr as shown on the charging screen. Here the vehicle range is the area under each curve. The area under all three green lines is the same; as is the area under both blue lines.
Finally, we get to the most useful graph: range vs time. Specifically EPA combined range gained over time starting at 0% SoC and ending at 100%. This chart makes it apparent the e-tron on a 160kW charger effectively charges like a 3 LRD/P on an Urban Supercharger. It also highlights the difference in available range.
Given a starting range, someone can trace over to the time and see how much additional time is needed to reach a desired rated range. For example, someone with a Model 3 showing 100mi remaining and who needs 150mi more can see that the charge session would take about 20 minutes on a V3 Supercharger, under optimal conditions. That's calculated by subtracting the 7.5 minutes shown at 100mi from 27 minutes at 250mi. Similarly, it's
also 20 minutes on a V2 Supercharger and 31.5 minutes on an Urban Supercharger. The e-tron takes...well, it doesn't go 250 mi.
So again I'll emphasize the Model 3 curves are only representative and need to be calibrated with more data from V3 and V2 charging sessions using the latest firmware. Also, these curves depict, by intent, the rates under optimal conditions. Cold batteries and warm chargers will reduce the charge rates. I hope this shows why I'm personally only interested in the first graph: kW vs %SoC. It will lead to the intermediate and final graphs that will yield many conclusions that are worthy of other discussions and threads. For now, I'm most interested in more people showing us what happens at Fremont!