Model 3

Temps, Precondition, Regen data with GRAPHS

1356710

Comments

  • edited November 2019
    The other timestamps that are identical included other data from other sensors?

    I understand.

    If you took each sample time and subtracted the previous sample time you would see how much jitter there is in sampling time. If data came every 1 msec, then the difference would always be 1. In data above the sample time differences are 1, 12, 0, 18, and 9. Lots of jitter. It is what it is. No worries.
  • edited November 2019
    Yes lots of jitter, but the theory was that the noise was caused by jitter. After doing a quick jitter analysis, the jitter is "even" throughout the dataset. The graph shows the "noise" reducing at a certain point, but there is no difference in the jitter distribution throughout the entire dataset. So I don't know why it is noisy and then levels out, but in my opinion it is not because of the jitter profile.
  • edited November 2019
    Detoram: sorry I'm late to the data party. Re: 1M row limit in Excel: have you tried the powerpivot add-in? Comes with all the recent versions, and can handle 10M rows easily, more I'd you have 64b Office.
    Really makes plowing through data this size much simpler and faster.
  • edited November 2019
    Thanks for checking that. So that rules out that it was a sampling jitter problem.

    I'm sticking with my idea that they ramp up the heater somehow and while ramping it shows up noisy. But when it hits steady state final "all on" the noise goes away.

    Plus the "noise" actually doesn't look like noise. 1) It gets larger as the actual power gets larger. 2) It's hard to tell without zooming in, but it looks periodic to me, not random.

    Typically, signal to noise ratio gets better as the signal gets larger because the noise stays the same amplitude.
    In this case the SNR is staying the same because the noise is proportional to the signal.

    What does the power output look like for the first minute of a car starting up, but without letting the heater start?
    That baseline would help us to see if any other component plays a role in the power.
  • edited November 2019
    The same noise is there for the super quick ramp down of the heater at the end of the graph.
  • edited November 2019
    On your 10 min of BP while not in motion graph, did you turn BP off at the end of the graph? Or was it starting to cycle? I know it is just rear motor, but it never gets up to the 7kW that someone else has reported. Looks like it is more like the 4kW we know and love.
  • edited November 2019
    Detoram: re: missing values. In powerpivot you can create a time series table, create a relationship between your time series table and your data table, then use the time series as your x axis.
  • edited November 2019
    @jallred: define startup...batt power is around .5kW when you get in the car...even stepping on the brake doesn't do really anything to that.
  • edited November 2019
    @majassow: I will have to look in to that, thanks
  • edited November 2019
    @jallred: yes I turned off the heater at the end. It leveled out at about 7kW in the graph.
  • edited November -1
    @jallred: just got in my car... .25kW, stepped on the brake and it went to .40kW.
  • edited November 2019
    So in the cabin heat data set all of the power is going to the cabin heater (except for .4kW). They don't just turn the heater on full, they ramp it up over about a minute. And while they are ramping it the power is somewhat noisy and the noise level is proportional to the actual power. When they get to full power (6kW) the noise goes away. When the heater gets turned off, it ramps down (but within a couple seconds) and it has similar noise during those couple seconds.

    During preconditioning, the stators are being heated and it consumes about 4kW on the rear motor.

    When the car is moving and preconditioning, heat is directed from the motor to the battery, but the motor seems to be in normal driving mode. No additional power is used for preconditioning while driving. At least according to the data you have collected.
  • edited November -1
    Looks like you got it...all the way till you got to "No additional power is used for preconditioning while driving."

    The jury is still out on that. It's a hard thing to test but I do know from looking at the data live that it does add a heating power waveform to the motor at low speeds. I would need a dyno to be able to get a good answer.
  • edited November 2019
    My empiric test suggested about 115 Wh/m were attributable to battery heating in motion. The battery heating will go off at about 32C. The Reddit experiment showed it go off after a couple kWh and a temperature rise slope of 1C per 132 Wh, from memory.
  • edited November 2019
    Yes. I think you are right under most conditions. Although there may be some conditions where it is happen with heat generated by normal driving profile. Bottom line is that worst case power is when stationary. It only gets lower when moving.
  • edited November 2019
    Yes, now on the AWD while driving you will get heating pwr to the front motor which isn't used a lot in normal driving conditions.
  • edited November 2019
    Very interesting. Strange Mr F is absent. I guess he is not interested in energy consumption anymore and his winter strategy. LoL.
  • edited November 2019
    Mr F is a very wide net :)
  • edited November -1
    Added Hold mode vs Roll mode Regen graph
  • edited November -1
    Nice new data. Thanks again for your efforts.
  • edited November 2019
    No problem, I love playing with it and I was wanting to get hard data to counter some of the crap info some people have tried to post.
  • edited January 7
    Added a new data point IRT cabin heater.
  • edited January 9
    @derotam, great post, I don't know how I missed this earlier. This will take longer than my lunch break to digest.

    Any case, just to understand the setup better...

    How far away was the SC you were routing to and what charge rate does it support?

    I've assumed the in route battery warming takes into account distance to avoid premature warming. Also, I would assume warming would be a function of the max charge rate of the SC you are targeting as charging at higher rates would require a warmer battery.

    Do you notice anything along either of these lines?
  • edited November -1
    @ADinM3: The supercharger I was using was approx 15 miles away. Supercharger charge capability is mostly irrelevant, unless the battery target temp follows a curve based on SC capability. I think that is a stretch right now and there is no indication that that is true. I don't think there is a target temp currently enumerated in the ScanMyTesla app and I don't feel like sitting around and wasting the power to find where preconditioning turns off.

    As far as distance, I really think there is just going to be a max time limit for when preconditioning would start. IT would not be good for the public to see preconditioning turn on right at the beginning of a 200 mile leg to the next supercharger. That is just my opinion though.

    Yeah there could be a curve fit on the target temp based on supercharger capability but that would be a little difficult to test as there isn't currently a battery target temp enumerated in the ScanMyTesla app.
  • edited January 9
    @derotam: Nice data! Thanks for posting this.

    I particularly like the Regen vs. Time with speed data as you can use that to estimate the regen efficiency by integrating kW over time -- area under the green curve -- and compare to delta Ke of the vehicle going from 70mph to 0.

    My measurement of the Regen energy from this graph is 241Wh -- interestingly, about 1 mile of range.

    Assuming your LR RWD had 300lbs of driver + cargo, the kinetic energy of the car traveling at 70mph is 915.6kJ or 254Wh.

    Which results in a regen efficiency of 95%. This matches my empirical calculations I posted about a year ago.

    If you have a better estimate of the total cargo + driver weight, I can refine that number. Each 100lbs represents a 3% change in the efficiency calculation.
Sign In or Register to comment.