Why are the power curves so delayed? For example, at 1mph, the cars show it's just at minimal power, like 40kwh. My guess is that it's a trailing average. It seems like the peak power output numbers would be the ones to trust. Is that correct?
There is some trailing average effect but look at the CAN data for Bill D and MikeBur which is more than a magnitude more frequent. The car doesn't reach full power when you press the accelerator. The traction control never comes on to limit the tire spin which should happen on every car. Something else is limiting power, both to differentiate the cars as well as to protect on the massive forces generated on the parts during initial acceleration.
I believe there are four primary drivers of performance and differentiation in the cars: Torque limiting, Power limiting, Current limiting and RPM limiting. In that order. Secondary factors are tire grip, thermal constraints of the individual parts and the electrical wiring. The shape of the power output over time supports this theory.
First, the initial rate of change of power (power acceleration) in constant. It is likely that the car is either set that way in the throttle mapping or that there is another setting (like torque) that results in a linear rate of change or power. This is true until the max power setting is reached.
Second, the max power is reached and the power is relatively constant through the 0-60 time on the slower cars.
Third, the power starts to drop as current (ampere) limits are reached. On some of the faster cars it starts to dip at the end on the 60 mph runs but certainly within the quarter mile you can see that the car power has an inverted, downward sloping curve. I believe that this is due to the exponential amounts of current required to offset the drag of the vehicle through the air at higher velocities, and consequently the remaining current available not being enough to reach max power.
Finally you hit the rpm limits of the motor and the top speed allowed.
@Krash, I need to challenge what you are calling "power acceleration" in your charts, to which you attach units of kW/s^2. Where did you come up with such this concept? I can understand how one might be interested in the rate at which power increases, whose rate of increase would be measured in kW/s, and would be equal to the slope of the kW vs. time graph (which is a mostly a flat rate of increase in the plots I have seen). However, I cannot image that the acceleration of power is a meaningful or measured quantity (and would be zero if the kW/s rate value is mostly constant.
SucreTease I welcome your peer review. Rate of change of power IS zero once the power limits are reached. My point in measuring it is that the initial rate of change of power is such an obvious performance differentiator between sport (corked), insane (uncorked) and ludicrous cars at approximately 75, 150 and 300 kWs2. It is the most noticed difference in the cars by the drivers, much more so than max power and the current limits, which are the other settings changed in uncorking. (rate of change of power, max power and max current being the major performance differences between Tesla cars disregarding max battery and launch mode).
Again, I think the linear rate of change of power is determined by a fixed torque setting, which everyone would love to know. Linear rate of change of power is certainly coincidental with a constant torque as observed in Bill D and MikeBur CAN collected data. You will notice that we started posting our torque estimates based on the rate of change of power. Unfortunately there is no physics behind the torque estimates. We are merely fitting the observed data to the Tesla reported data and adjusting as we get more data or a real physics model.
Looking at the runlogs I see that the time starts at "negative" (-0.25 sek) for example.
Why is that? And why isn't that added to the total 0-60-time?
The timed metrics include a rollout. You'll see in the summary footnote that you should add in a quarter second if you don't want a rollout. But actually to be accurate you need to look at whether the end of the run was at 60 vs 63 and adjust the end time accordingly, as well as adjust the start time to interpolate when the accelerator was hit.
The main reason for the rollout is that for MANY of the forum supplied cars' runlogs the cars have rolling starts. I needed a time to set at zero. Using the rollout model of saying that the first >1ft distance measured is the zero time makes the API data pretty consistent and is close to the spirit of the magazine rollout model.
You will see even longer negative number leads on the cars that have launch mode since the car is actually loading power on the brakes in advance of the release of the vehicle.
The fact is I am not interested in the 0-60 times or quarter mile times. We have some track verified times and I haven't included them, although on reflection it would be interesting to estimate them from this data. I do think the components that determine those times are interesting because they are all components that Tesla and owners can improve independently over time and because the components have their own practical implications and constraints.