I think it only takes into account temperature when you leave. So if your car is nice and toasty in the garage, then you go outside into -15 that may be throwing it off. An interesting test would be to start a trip and note the destination %, then once your car shows the correct outside temperature recalculate the trip and see if it estimates a lower %. Temps are to warm for me right now to test this effectively.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Evaluation and Observations with v6.1 Trip Energy Prediction
- Thread starter Vger
- Start date
-
- Tags
- Model S User Interface
I think it only takes into account temperature when you leave. So if your car is nice and toasty in the garage, then you go outside into -15 that may be throwing it off. An interesting test would be to start a trip and note the destination %, then once your car shows the correct outside temperature recalculate the trip and see if it estimates a lower %. Temps are to warm for me right now to test this effectively.
gzerninplatz
Member
I believe that it includes altitude, but NOT temperature. You have to fudge around that. It obviously is easier to include altitude - which doesn't change - than temp, which constantly is changing over time for every location...
You are right it may not take into account temperature at all, and just current HVAC draw.
It does. It accurately accounts for your local temperature, as long as you live in a greenhouse set to 22C/72F.
In the graph you posted, you actually did pretty well, only 15%-20% over the predicted consumption. I overeat by 20%-30%, but perhaps it's a bit colder here. And yes, my tires are inflated to be 45 psi at -17C. I'm worried they might pop when in a heated garage.
No, seriously, the 6.1 estimation is very close to the one you get in evtripplanner when using the default exterior and cabin temps of 22C. The evtripplanner predictions when entering the correct -15C to -20C for the exterior temperature are much more in line with reality.
mknox
Well-Known Member
It was a 39.4 mile trip and the energy consumption was 419 Wh/mi. That's actually not bad for this time of year. In the summer I can get close to, and even beat 300 Wh/mi.
- - - Updated - - -
Interesting. It was probably about +10 F in my garage. It's attached to the house, but not heated. Still pretty cold, though. Others are suggesting it might not take temperature into account, so that could be a likely culprit too.
1) Driving style is big determinant, as well as characteristics of route and weather. In dead of winter, for 40 mile trip which includes about 50% highway, flat Chicago (like Toronto), I'll ALWAYS beat 400... In summer, that would be around 280. So, I suspect you may enjoy the, er, sporting characteristics of the MS more than me:smile:
2) Conjecture I've read is that it uses level of HVAC use as proxy for temperature. Same effect... if you initiate trip while car is still warmed from your garage, you are probably using less HVAC.
IMO, temperature should be much more easy to factor into the calculation than wind speed and direction - the latter is constantly changing. Temperature sees much less variation over a given 2-3 hour period that you will be driving. All the Nav has to do would be to look up the forecast for waypoints along the route and determine if excessive HVAC use or battery conditioning would be necessary.
I have tested this and the estimation does not change when you move from a garage heated to 60°F to outside where it's 0°F.
With the vehicle outside for over 30 minutes in 0°F and the heat running, 6.1 estimated that I would use 69% of the battery for a 244 km trip. Then when I canceled the destination and turned off the heat, reentering the destination led to a guess of 65%.
This is astonishing. Just 4% of a full charge (~3 kWh) to heat the vehicle for almost three hours in such bitter cold. Hats off to the engineers for designing the system to so effectively reuse waste heat to warm the cabin. However, I suspect this is too good to be true when it's this far below freezing. Perhaps the estimator doesn't take into account the waste heat being diverted to warm the battery when it's so cold.
Edit: Whereas the estimator in 6.1 considered the heat would only use ~3 kWh, evtripplanner said 11.2 kWh. (the other differences are that evtripplanner said the trip would be 1.5 km longer, but take 30 minutes less to complete).
Last edited:
mknox
Well-Known Member
Quite honestly, I am generally the slow-poke over in the right lane. And during my "test" above, I was using every hyper-mileing trick I could think of.
I did couple trips to Minneapolis - both ~180 miles (90 miles each way). First trip was on a single day; prediction was within 1-2% at temps of ~30F and speeds ~68 mph, and I made it comfortably back home. Overall energy use 373 Wh/mi.
Following week I had to do the same trip at 0F. The software predicted arriving at 49% in Minneapolis, but then dropped steadily for first 15-20 minutes down to 34%, then stayed constant for rest of the trip (I drove 65 mph). I knew the parking next to the hotel had a ChargePoint, so I did not worry. I used ~ 480 Wh/mi.
Next day on the return home it was still ~0F, but otherwise gorgeous: bright sunny day, no wind at all. Same story - predicted range dropping quickly first 15-20 minutes, then steady, with overall 10% less than predicted (residual 22% instead of 32%). Return trip 435 Wh/mi despite going 74 mph. I suspect the difference was headwind the evening before.
My conclusions on P85D trip calculations (.139 version):
1. Pretty much spot on for temps ~30F
2. Count ~10-15% absolute extra use for each 100 mile at ~0F (i.e., 50% predicted residual ends up being only ~ 35% residual level).
Great cold weather test this am. 79 mi trip, 50% hwy 30% rural, 20% city. -20c. Yes, that cold. Avg 397 wh/mi. Predictor said 59% battery @ destination, which it revised to 56% within a few miles from start. Ended with 57%. It has to take into account temp somehow.
PS: my 21k A battery charged to 265 for this trip. Must have won the battery lottery!
PS: my 21k A battery charged to 265 for this trip. Must have won the battery lottery!
Cottonwood
Roadster#433, Model S#S37
I have noticed an interesting anomaly in the energy prediction when going through tunnels. See the screen shot of my going from Silverthorne to Boulder yesterday and note the circled area which was me driving through the Eisenhower Tunnel..
I could not figure out why the predicted energy had such an inverted spike, and used less than reality. My current theory is that the prediction model is going over the top, instead of through the tunnel. That would be what it's topographical database would have. My guess for the reduced total energy used, is that the model thinks it is going off-road at low speed.
Thoughts?
I could not figure out why the predicted energy had such an inverted spike, and used less than reality. My current theory is that the prediction model is going over the top, instead of through the tunnel. That would be what it's topographical database would have. My guess for the reduced total energy used, is that the model thinks it is going off-road at low speed.
Thoughts?
billarnett
Member
I've noticed that sometimes the navigation software will get confused and want me to get off the highway and onto the frontage road/side streets - and then get back on the highway at a subsequent ramp. If the nav software believes you'll be driving at a lower speed for that period, that would produce the dip in energy consumption.
Cottonwood
Roadster#433, Model S#S37
Maybe it thought the route was over the Loveland Pass Rd? (Isn't that more fun anyway
Nope, the map had me going through the tunnel, not over Loveland Pass, and there is no other surface street. I think the topo lookup just does not understand tunnels, yet...
I think your analysis is correct. The slope of the line goes twice as steep as it is for the grade up from Silverthorne. Maybe after Stevezzzz is done with his coasting experiment down the hill, you could confirm this by taking your P85D over the top. The Hakkas should help. Naturally I will stick around to film this for posterity.
BTW, you did a very good job of beating the curve all the way down the hill. What were the driving conditions? (If you say "traffic jam from all the skiers", I will not try to replicate your numbers.)
mspohr
Well-Known Member
I've found the the initial prediction is always more pessimistic than actual results. It does adjust frequently during the trip so it quickly moves into a more realistic range. It might be just the way I am driving. I'd hate to see it go the other way.
Also, on a recent trip through Death Valley, I found the prediction had a hard time with the mountain passes and deep valleys (three passes 5000-6000 feet and valleys to below sea level. On this trip it gave me estimates which moved from +10% to -23% and back as I would go up and down the steep mountains.
Also, on a recent trip through Death Valley, I found the prediction had a hard time with the mountain passes and deep valleys (three passes 5000-6000 feet and valleys to below sea level. On this trip it gave me estimates which moved from +10% to -23% and back as I would go up and down the steep mountains.
The trip energy calculator in earlier versions of 6.1 was overly optimistic about predicted energy usage during climbs and pessimistic about usage during descents. Cottonwood reported that these errors have been at least partially corrected under .179.
Cottonwood
Roadster#433, Model S#S37
I've skied that hill from the top to abreast the tunnel entrance at Loveland. I have no interest in driving that in anything other than a snow cat. Every snow cat I have seen is diesel powered...
I was just doing normal PSL+4 driving. The only big difference is that almost half of the route from the Eisenhower Tunnel to Golden has the PSL temporarily reduced from 65 to 55 because of construction. Traffic was pretty light.
Similar threads
- Article
