That's funny! When I moved to Pocatello many years ago, I was told that there was Winter and the Fourth of July. Then they through in the caveat that one year, the fireworks on the Fourth were canceled. Because of snow.....
Analysis of the Tesla X curve here shows that it should consume about 300 Wh/mi at 70 mph which in fact in my couple of years driving one is very close to what it does. What is disturbing about this curve is that it indicates that out of that 300 Wh/mi only 93 Wh/mi are attributable to drag. That seems low.While manufacturers can't really sandbag the tests themselves, they do provide the numbers (aerodynamic drag, tire rolling resistance, and driveline friction) that set the road load on the dyno. Teslas numbers reflect a much lower road load than the Taycan. Tesla seems to be more aggressive in their road load and Porsche much more conservative (Teslas will indeed have lower road load, but probably less than the dyno settings reflect)
An analyst looking at this data would say "Hey, there is no road load when the vehicle is standing still therefore the first coefficient has to be 0" and use that fact in doing his fit (calculation of the other two coefficients). I did that in looking at this curve but one always sees reference to 3 coefficients when this is being discussed. Can it be that a real analyst never looked at this?
For most automakers, the EPA takes the raw dynamometer mileage of the car tested using EPA protocols and multiplies it by 0.70 to approximate the effects of real-world road conditions on range. The EPA approved using a 0.75 factor for adjusting Tesla mileage based on Tesla's claims that its ancillary systems were more efficient than other EVs. I suspect the use of this more favorable factor in assigning a range number to Teslas is one of the reasons that Teslas invariably fall the shortest of their EPA-rated ranges of any brand in real-world driving. Their ancillary system efficiencies apparently aren't what they're cracked up to be.
As the article linked in No. 38 explains the dyno tests are run following SAE protocol. As it says there and in dozens of posts here and elsewhere there are 5 tests in the SAE suite. The EPA came up with a fudge factor, presumably that compares the 5 test set to the two test, which they allow for use in estimating EPA range if you only do the 2. If you do the 5 obviously you get a more robust result and so the adjustment factor doesn't have to be degraded to compensate for the missing data. Tesla and Audi do all 5 runs and so are permitted to omit this degradation. If the 3 additional tests push the composite higher they are allowed to indicate this.
This is upside down reasoning. The Tesla (and Audi) numbers are more robust than anyone else's and are better predictors of actual range than anyone else's. An EPA estimate that resembles realized performance at highway speeds is a terrible estimate. It is not supposed to represent performance at highway speed.
They are every bit as good as they are cracked up to be as indicated by their ability to give accurate range predictions under actual driving conditions.
The first coefficient doesn't have to be zero unless the test requires the fit curve to accurately model load at zero mph.
The main reason to use this physical fact is that in improves the "geometric dilution of precision" in the estimate of the linear term (b) appreciably.
The retarding force is modeled as: F = a + b*v + c*v*v
I have come up with a much better system than the EPA. Take 5 people age 50 and have them drive in a City at the posted speed limit for half a tank. Then have them drive on the Highway for the balance of the Tank at 70 MPH. That is what all of us will do and then we will be reporting the mileage we got. It would save millions in testing cost and we would not have to guess the mileage number. It would also be much more accurate.
I might be able to comment if I knew what 22 and 29 were.
People that understand what the EPA ratings are (a benchmark) know the wisdom of your proposal. What they do is drive the car and use its instrumentation and third party apps such as TeslaFi and Stats to learn how the car, as driven by them, deviates from the benchmark. If, for example, they observe that freeway consumption goes up from 300 to 380 Wh/mi on a rainy day they will know to expect a penalty of 80 Wh/mi if it starts to rain. Etc.
The vehicle is stationary for less than 0.1% of the highway test cycle, so precise modeling of stationary conditions is pretty trivial. It's entirely possible a small non-zero first coefficient more accurately describes the rest of the data.
Sorry, I meant what did you come up with for B and C. I didn't see a data table and was wondering if you were eyeballing it from the graphs.
It is not only theoretically true. It is actually true. And so, if a quadratic is a good model, and it should be, then using the zero point improves the DOP (dilution of precision appreciably. If the quadratic is a good model only in some region which excludes the origin then it's a different matter. Should that be the case then I have a problem doing what I did because the data in those curves doesn't have enough span to estimate the coefficients from it (the DOP is poor).
Well the curves didn't come from a peer reviewed journal. They came from an article in a motoring magazine written by journalists - not scientists or engineers - so one needs to take them with a grain of salt.
Not sure how much detail you want but in case you want the whole 9 yards here it is. Yes, I eyeballed points off the graph. Here's what I got:
While the EPA test might not be meant to indicate highway range accurately, it is highway range when driving on long trips in real-world conditions that is of most concern to EV owners. Range is far less critical for local driving for most EV owners, especially when home charging is used.
