Why does my Model 3 use significantly less energy in self-driving mode?

[strykeroz]

Time to admit the truth... it's a better driver than you.

FWIW on highway I admit it is in my case. On a recent trip to the coast I was getting 110Wh/km at 110km/h with 2 people and some gear on board with AP. That's the kind of numbers I get on my daily commute, on my own in an empty car, and only if I'm being frugal (ie not that often if I'm honest) at 60km/hr.

 

[KG2V]

I tend to be one of those go with the flow, middle/right lane drivers. Last year (this year I wok in the city - traffic) when commuting by highway, I regularly was able to come home at an average of less than 190Wh/mile, and it was rare to be over 200. The answer is "be smooth".
I used to watch the NY State officers drive, and they were always so smooth (they don't seem as smooth today). I always tried to emulate that. You don't have to be SLOW, but SMOOTH.
Get to the point where for long portions of the trip you actually see NO bar, green or black - the car just rolling along (MUCH harder to do with single peddle mode)
"Slow is smooth, and smooth is fast"

 

[jkirkwood001]

The OP said he was mimicking AP as best as possible. Which might well mean more regen than necessary. But it should be easily possible to drive more efficiently than TACC. Perhaps not over the long term, but certainly if you are trying to drive efficiently it should be easy to do much better than TACC (but I doubt it would significantly impact efficiency due to diminishing returns - TACC is fairly efficient).

I agree. All cruise controls will "step on it" to maintain speed going up hill. This always seemed wasteful to me. Can't I have an intelligent AutoPilot that keeps the same energy consumption - within reason - no matter how hilly? Downhills are an exception - don't go over my set speed. I assume trying to maintain 120 KPH / 70 MPH uphill wastes many KM / miles of equivalent flat driving. And AP is never so smart to accelerate a bit before the hill and back off while climbing.

Thoughts?

 

[KaptKamp]

My guess is that almost any kind of cruise control saves energy on both ICE and EV cars because your speed is constant. Especially at the end of long trips your instinct is to drive faster due to being tired and wanting to finish the drive. I can't imagine it would be as much as 10-20% unless you are varying your speed on manual over a wide range constantly. Personally I have not noticed any particular energy savings in Autopilot though I do arrive at my destination feeling less tired.

From my personal experience in ICE cars and more limited experience in my new Model 3, cruise control is always LESS efficient than human feel. Although speed is more constant, engine power is not. Driving on the highway you will find that although your pedal movement is minimal, your speed fluctuates up and down with factors like incline, wind, drafting with a convoy of cars/trucks. Also, we consciously (or subconsciously) anticipate situations, lifting off early when you feel a car might cut you off from a slower lane or accelerating up behind someone early with their blinker on as they begin to change lanes. The car is only able to react to definite situations, only braking once the car has fully come into your lane, or accelerating once they are fully out of the lane.

 

[Daniel in SD]

I agree. All cruise controls will "step on it" to maintain speed going up hill. This always seemed wasteful to me. Can't I have an intelligent AutoPilot that keeps the same energy consumption - within reason - no matter how hilly? Downhills are an exception - don't go over my set speed. I assume trying to maintain 120 KPH / 70 MPH uphill wastes many KM / miles of equivalent flat driving. And AP is never so smart to accelerate a bit before the hill and back off while climbing.

Thoughts?

Aerodynamic losses go up with the square of speed so for example it’s better to go 65mph for an hour than 60mph for half an hour and 70mph for half an hour.
So what’s left is motor efficiency. The motor actually gets more efficient at higher loads so that would suggest that you should actually go faster up hill. On the other hand it gets less efficient at lower loads and regen is super inefficient. My guess is for optimal efficiency you would want to go fast enough on the downhills to never use regen and then as slow as you like on the uphills to reach your target average speed.
I prefer to go faster on uphills since braking distance is shorter and there are less likely to be speed traps.

 

[M_D129]

Yep, no surprises here. How do you know TACC is using the brakes? Small difference but there and particular types of driving might bring it out a bit more. Still, really really hard to measure small differences like this. I’d expect less than a 20Wh/mi difference with conservative driving with the same speed profile.

I rest my foot gently on the pedal of the brake, in the event I need to react, more often than not, I witness very little regen, in the status bar, and feel the brake depress. Stop and go traffic is one of the instances I see this the most. When I am operating, I take the most advantage of regen, in stop and go, and very rarely have to apply the brakes.

 

[Paddy3101]

My wife uses about 50% more energy than me. Not any faster, just uses more energy. She is constantly micro-adjusting the speed throttle, not enough to make any difference to the speed, and mostly you can't feel it (or I have just got used to it), but you can see it on the black/green bars. Constantly up/down/up/down. Black/Green/Black/Green. I'm usually around 260-280, wife is 350-400, but same average speed.

Yesterday on one 8 mile trip she averaged 441wh/m with an average speed of 17 and a max of 67. That was quite impressively in-efficient! 63 degrees and sunny!

Regen is better than using the friction brakes, BUT still much worse than not accelerating, only to slow down again.

Don't over-drive the car or the traffic, let it roll. You'll be quicker AND use less wH. Takes practice.

Take notice of what big trucks do, they don't have the energy output to constantly accelerate/decelerate AND they want to minimize their costs. Leave a gap and coast/use the gap as a buffer against having to brake/accelerate.

 

[Fresnel]

If you're driving efficiently on freeways you really shouldn't be using regen or friction brakes.

You will regen on declines / going down hills. Unless a person is accelerating down hill

 

[Paddy3101]

You will regen on declines / going down hills. Unless a person is accelerating down hill

Only if it's steep enough for gravity to be greater than the drag of the car. At freeway speeds, the drag is quite high, so the hill needs to be steep(er).

The faster you go, the more drag, the less re-gen needed. So let it roll, till it reaches an equilibrium!

You then have captured 100% of the energy from the hill (that's available) to be repaid going up the next hill, rather than 70% (made up number) storing the excess in the battery.

 

[Fresnel]

Only if it's steep enough for gravity to be greater than the drag of the car. At freeway speeds, the drag is quite high, so the hill needs to be steep(er).

The faster you go, the more drag, the less re-gen needed. So let it roll, till it reaches an equilibrium!

You then have captured 100% of the energy from the hill (that's available) to be repaid going up the next hill, rather than 70% (made up number) storing the excess in the battery.

Exactly

 

[JPWhite]

My Model 3 (dual motor long range AWD, Fall 2018) uses significantly less energy when it is on autopilot than when i drive the same roads at the same speed myself, as gently as possible, and using the single pedal (in standard mode)? On autopilot, which I use only on freeways or well-marked roads, it gets the rated mileage (245-250 Wh/mile, leading to a 310-mile range). I can drive the same roads myself, going almost exactly the same speeds, trying to mimic the autopilot's driving as closely as I can, and I do 10-20% worse. Does anyone know the reason for this difference? Does the autopilot route different amounts of power to the two motors as a function of speed and incline differently from what can be achieved with the accelerator pedal?

I have found the opposite to be true. AP is a heavy braker compared to me.

 

[camalaio]

Smoothness is absolutely key.

You probably gain nothing from coasting downhill to a higher speed. That higher speed causes significantly more drag, and you're losing energy to that more than you would if you kept a steady speed. Or that's my theory anyways.

Edit: this assumes you're using regen. Coasting is absolutely better for an ICE vehicle. Higher speed without input throttle is always better in their case.

Speed steadiness is likely the difference for people who get better efficiency with cruise control. It doesn't really overshoot the target speed, however a human often does by a little bit. That little bit is disproportionately worse in terms of aerodynamic drag, so you actually lose a fair bit of energy over time.

That said, cruise control isn't as smooth as you can get. Drivers can definitely do better than AP.

I agree. All cruise controls will "step on it" to maintain speed going up hill. This always seemed wasteful to me. Can't I have an intelligent AutoPilot that keeps the same energy consumption - within reason - no matter how hilly? Downhills are an exception - don't go over my set speed. I assume trying to maintain 120 KPH / 70 MPH uphill wastes many KM / miles of equivalent flat driving. And AP is never so smart to accelerate a bit before the hill and back off while climbing.

Thoughts?

My thought is keeping the same power to the motor as if you were going 120km/h but up a steep hill would maybe result in a steady state speed of 30km/h. You have a very heavy car that you're hauling up a hill, it requires insane amounts of power to do so at a reasonable speed.

Try keeping the black bar constant some time prior and during an uphill. Notice speed is likely lost quite rapidly.

My experience with cruise in a Model 3 is that it is rock solid, very rarely fluctuating by +-1km/h. Most gas vehicles are significantly worse, especially with the lag and momentum of the drivetrain.

I rest my foot gently on the pedal of the brake, in the event I need to react, more often than not, I witness very little regen, in the status bar, and feel the brake depress. Stop and go traffic is one of the instances I see this the most. When I am operating, I take the most advantage of regen, in stop and go, and very rarely have to apply the brakes.

In stop and go and likely especially with the Roll stop mode, yeah TACC/AP tends to use friction brakes a lot, especially since its overall not very smooth in such traffic.

 

[Daniel in SD]

In traffic AP just follows the car in front of you. If you're a less efficient than average driver then AP will be better than you.

 

[Daniel in SD]

You probably gain nothing from coasting downhill to a higher speed. That higher speed causes significantly more drag, and you're losing energy to that more than you would if you kept a steady speed. Or that's my theory anyways.

It does if the alternative is using regen. The optimal speed going downhill is higher if the hill is steep enough to use regen.

My thought is keeping the same power to the motor as if you were going 120km/h but up a steep hill would maybe result in a steady state speed of 30km/h. You have a very heavy car that you're hauling up a hill, it requires insane amounts of power to do so at a reasonable speed.

Theoretically it takes the same amount of energy to go up the hill fast or slow (neglecting aerodynamic losses). If you go slower up the hill that means you'll have to go faster later to achieve the same average speed and that will result in more total aerodynamic losses.

 

[Fresnel]

Just look at the big green regen bar when going down a big hill... don't need to over think this.

 

[camalaio]

Theoretically it takes the same amount of energy to go up the hill fast or slow (neglecting aerodynamic losses). If you go slower up the hill that means you'll have to go faster later to achieve the same average speed and that will result in more total aerodynamic losses.

Same energy (kWh), yes. Same power (kW, hp), no. Think of it this way. If it took 10 power units continuously to climb a hill no matter what the speed, you'd have different amounts of time. Let's say 10 power units over 10 minutes is 100 energy units, but if you go slower and take 30 minutes, it's 300 energy units. Therefore you cannot use the same power for different speeds (ignoring resistance) and end up with using equal amounts of energy.

 

[Daniel in SD]

Same energy (kWh), yes. Same power (kW, hp), no. Think of it this way. If it took 10 power units continuously to climb a hill no matter what the speed, you'd have different amounts of time. Let's say 10 power units over 10 minutes is 100 energy units, but if you go slower and take 30 minutes, it's 300 energy units. Therefore you cannot use the same power for different speeds (ignoring resistance) and end up with using equal amounts of energy.

Yeah that's what I said, it takes the same amount energy. We're talking about energy per distance right?
Obviously if you want to minimize power you should just keep the car in park and asleep.

 

[Fresnel]

Same energy (kWh), yes. Same power (kW, hp), no. Think of it this way. If it took 10 power units continuously to climb a hill no matter what the speed, you'd have different amounts of time. Let's say 10 power units over 10 minutes is 100 energy units, but if you go slower and take 30 minutes, it's 300 energy units. Therefore you cannot use the same power for different speeds (ignoring resistance) and end up with using equal amounts of energy.

Not true because faster speed would use more energy unites than slower speed. Much like how driving 75mph uses more than 60mph.

 

[Daniel in SD]

Not true because faster speed would use more energy unites than slower speed. Much like how driving 75mph uses more than 60mph.

(neglecting aerodynamic losses)

(ignoring resistance)

 

[Fresnel]

And what's the point of ignoring them?