I finally got a minute to do some analysis on the supercharging data I collected during our cross-country trip (blog here -- ElectricStartupSuperTrip pseudo-blog thread!) and wanted to share it.
Attached is an excel file that contains VisibleTesla data from the ~30 supercharging stops we made along the way. I'll try to give a brief overview so people know what they're looking at...
The original point of this particular exercise was to answer the question I asked here: Model S charge curve as a .CSV or easily readable graph?
I also added a couple of scatter plots: the "miles per minute per charge" has all the data points, so it's a little noisy. The noisy plots are things like the very first data point where the charger hasn't spooled up to full power, or where there are gaps in the data, or similar crap like that. There's also just enough variance in how the time deltas are calculated to introduce some noise.
The "charge rate v. rated miles" plot is one where I cleaned up the data by doing things like eliminating the first data point from each supercharging session - to get rid of the initial ramp up error, and then eliminated all values past 240 miles of rated charge since things really really slow down then, etc. But when you just plot the computed charge power (i.e. what's actually going into the battery) vs the state of charge, you get a really really consistent curve that declines as the battery fills up and fits extremely close to the expected values. To sanity check these, I used the data to build a little calculator to answer the question: "If I roll up to a supercharge with XXX miles on the dash, how long will it take to get to YYY miles"? That's the "Timer" spreadsheet. I assumed that when you plug in it takes about 2 minutes for the car to get "spun up" and after that the charge rate tracks the data in the spreadsheet. I spot checked these times with the actual data from my supercharging stops, and it looks really good. To use the timer, just subtract the starting value in minutes from the ending value in minutes. So to go from 40 miles of range to 140 miles of range -- 28-9 = 19 minutes.
IMPORTANT OBSERVATION: the "reported" rate of charge (col G), or at least whatever VisibleTesla calls that column, has a ton of error in it. I looked at that data for a long time and could not - for the life of me - figure out what it correlated to. If anyone has thoughts on this, I would LOVE to hear them. I plotted this against the battery "fullness" in the "charge rate v. SoC" tab. The data looks like crap, and when you do the math it doesn't correlate to anything. Not realizing this data was bogus cost me a TON of time. Grrrrrr tesla! (I'm assuming that this is an error in the Tesla reporting HTML, and not a VisibleTesla computation error?)
I hope this is helpful to others!
Attached is an excel file that contains VisibleTesla data from the ~30 supercharging stops we made along the way. I'll try to give a brief overview so people know what they're looking at...
- Column A is the date.
- B is the rated miles of range reported remaining in the battery
- C is the reported current into the battery during charging
- D is the reported charge voltage
- E is a COMPUTED value (more on this later) that just multiplies current and voltage - unit is kW.
- F is reported state (percentage) of charge.
- G is the REPORTED "RoC" from Visible tesla. (More on this later!)
- H is the amount of time (minutes) between two subsequent rows in the table.
- I is the COMPUTED rate of charge - I take the delta in reported range, divide by the delta in time. Unit: miles per minute of added charge.
The original point of this particular exercise was to answer the question I asked here: Model S charge curve as a .CSV or easily readable graph?
I also added a couple of scatter plots: the "miles per minute per charge" has all the data points, so it's a little noisy. The noisy plots are things like the very first data point where the charger hasn't spooled up to full power, or where there are gaps in the data, or similar crap like that. There's also just enough variance in how the time deltas are calculated to introduce some noise.
The "charge rate v. rated miles" plot is one where I cleaned up the data by doing things like eliminating the first data point from each supercharging session - to get rid of the initial ramp up error, and then eliminated all values past 240 miles of rated charge since things really really slow down then, etc. But when you just plot the computed charge power (i.e. what's actually going into the battery) vs the state of charge, you get a really really consistent curve that declines as the battery fills up and fits extremely close to the expected values. To sanity check these, I used the data to build a little calculator to answer the question: "If I roll up to a supercharge with XXX miles on the dash, how long will it take to get to YYY miles"? That's the "Timer" spreadsheet. I assumed that when you plug in it takes about 2 minutes for the car to get "spun up" and after that the charge rate tracks the data in the spreadsheet. I spot checked these times with the actual data from my supercharging stops, and it looks really good. To use the timer, just subtract the starting value in minutes from the ending value in minutes. So to go from 40 miles of range to 140 miles of range -- 28-9 = 19 minutes.
IMPORTANT OBSERVATION: the "reported" rate of charge (col G), or at least whatever VisibleTesla calls that column, has a ton of error in it. I looked at that data for a long time and could not - for the life of me - figure out what it correlated to. If anyone has thoughts on this, I would LOVE to hear them. I plotted this against the battery "fullness" in the "charge rate v. SoC" tab. The data looks like crap, and when you do the math it doesn't correlate to anything. Not realizing this data was bogus cost me a TON of time. Grrrrrr tesla! (I'm assuming that this is an error in the Tesla reporting HTML, and not a VisibleTesla computation error?)
I hope this is helpful to others!