Forums

Regenerative Braking

Regenerative Braking

Does anyone know if the regenerative brake will have an adjustable setting for more or less braking?
I am assuming that more braking increases the amount of energy recovered. City mode highway mode?

Mycroft | 8 januari 2012

No, we don't know if it will be adjustable or not.

ggr | 8 januari 2012

Roadster owners mostly don't want it to be adjustable. I'll be leaving mine on the maximum setting if it is.

Mycroft | 8 januari 2012

Agreed ggr. Personally, I would not make it adjustable. You very quickly get used to the regen braking. Adjustability might confuse people even more.

Timo | 8 januari 2012

There seem to be a misunderstanding (not saying that any of you have it) that regen is either on or off. That's not the case, it goes smoothly just like acceleration based on pedal position. That means that it really doesn't matter much how hard regen you get, you can control it with your right foot.

I would prefer strong regen over weak one (up to the point where tires start to slip), because it would allow me to drive more without touching the brake pedal. One foot driving.

dborn @nsw.au | 8 januari 2012

You get used to the regen within about 15 mins of driving, i found when i test drove the roadster. It was a really GREAT feature!!!

olanmills | 10 januari 2012

I don't want to be able to adjust it per se, but I would like the option for it to only engage when the brake pedal is pushed.

As it stands now, you can't coast. If you could coast, and if you drive in a more conscious manner, you could improve range. Applying the brakes on a car always wastes fuel/battery, regen brakes or not.

Timo | 10 januari 2012

Yes you can coast. Coast is between acceleration and regen.

Volker.Berlin | 10 januari 2012

This issue has been extensively discussed in some older threads, e.g.,
http://www.teslamotors.com/forum/forums/coasting-retarded-it-roadster

Maybe worth reading, before all arguments are repeated here.

ggr | 10 januari 2012

Timo, thanks for expressing it so clearly. Olanmills, I bet I can get better range by engaging the cruise control to a fixed speed on the Roadster, or even just using my foot to keep constant speed, than you could by manually switching to neutral (to achieve coasting). I'll even loan my roadster to the experiment, if you can come to San Diego.

olanmills | 10 januari 2012

You are right about keeping constant speed when you are just moving, but what about when you see a stop sign ahead?

You can drive the same speed and then apply the brakes very close to the stop sign and decelarate rapidly, like most people do. That's fine; that's a very normal way to drive.

Or you could coast and slow down more gradually and then apply the brakes with less force because you'll be going slower when you get closer to the stop sign. You'll save fuel this way.

Of course there is some balance. It doesn't make sense to coast at very low speeds for long distances, especially in an ICE car where you'll waste fuel.

Whether it's regen or physical brakes, it's braking. Braking always wastes energy* because you are trying to remove energy that you already put in the car to make it go forward.

*technically, it's wasting energy when you accelerate, it's too late to do anything about it by the time you need to brake. Basically, one way to look at braking is that when you're applying the brake, it's because you put too much energy into the car in the first place. You only wanted to go X distance, but you put enough energy in your car to go X+Y distance, and so you have to apply the brakes and get rid of the excess kinetic energy from your car.

Timo | 10 januari 2012

I bet you can't estimate the distance to stop sign so accurately with coasting that you don't need to apply some brake before you are at the stop sign. If you apply brake even a little, then you lose to regen by comparison.

I repeat my earlier statement: Coast is between acceleration and regen.

As for option to regen with brake pedal, I believe that can't be done by law. Law in many places force brake pedal to actually do braking.

Brian H | 10 januari 2012

Timo is not just being theoretical; it's the actual pedal position after easing off accel and before the regen kicks in. So it's even better than your guesstimating of the distance, etc. Some have suggested making a kind of "haptic" notch so that you can feel the gap between go&slow, but it doesn't seem to be functionally necessary.

Volker.Berlin | 11 januari 2012

As for option to regen with brake pedal, I believe that can't be done by law. Law in many places force brake pedal to actually do braking. (Timo)

I think that's precisely the way regen is implemented in the Prius (or was it yet another car?). There is no problem with the law, because if you press the pedal far enough, it'll apply regular brakes. It's just the first inch or so that decelerates the car by way of regen only.

With this specific car (whether it was the Prius or not) users reported that the experience was not as smooth and natural as it should be. In particular, regen is disengaged when actual brakes are applied. If the pedal is in a subtle spot between regen and regular brakes, it can actually happen that you press just a little harder, and the braking effect counter-intuitively decreases for a moment, instead of increasing linearly along the pedal travel.

However, this seems implementation specific and it occurs to me that if Tesla implemented regen in the brake pedal, the experience would probably be smooth, seamless and predictable. Still, with this approach, you cannot do the praised Tesla-style one-pedal-driving.

BMW seem to put quite a bit of thought into the regen implementation in the upcoming i3. They make a difference of "easing up on the pedal" and "easing off the pedal":

[...] The electric drive also allows for deceleration by means of the accelerator pedal. After the driver eases up on the accelerator, the electric motor acts as a generator, converting the kinetic energy into electricity which is then fed back into the battery. Energy recuperation generates a braking effect which makes a significant contribution to vehicle deceleration.

A coasting mode makes this unique "single-pedal control" of acceleration and braking using only the accelerator even more user-friendly. When the driver eases off the pedal, the electric motor's zero torque control keeps the drive train disconnected as long as the pedal is in this position. The vehicle now coasts without consuming power, driven by its own kinetic energy.
http://www.bmw-i-usa.com/en_us/bmw-i3/
(close to the bottom of the page)

Volker.Berlin | 11 januari 2012

And finally, there has already been an authoritative answer to the original question of this thread:

By luck, one of the first I introduce myself to is Drew Baglino, who happens to be directly involved in setting up the Model S regen braking system. [...] When I pleaded for driver-adjustable regen on the Model S, he said this was under consideration. (Two days after the event, while writing up this report, on a whim I e-mailed Tesla boss Elon Musk and repeated my plea for driver-adjustable regen. Three hours later, I got this reply: "I totally agree that regen should be driver-adjustable and it will be on Model S." The message was Cced to JB Straubel, Tesla CTO.)
http://www.thetruthaboutcars.com/2011/10/tesla-model-s-customer-blog-bet...

brianman | 11 januari 2012

@Volker.Berlin
Regarding the BMW quote...

1. Suppose you're travelling on a flat smooth road at 50mph at say 50% pressure on the accelerator.
2. You "ease up" on the accelerator to 25% pressure. I would expect the speed to start decaying...so you slow 45mph, 40mph, 35mph, ...
3. You "ease off" the accelerator (0% pressure). Because of the "drive train disconnection" the decay should be less pronounced: 35mph, 34mph, 33mph, ...

That's my interpretation of the quote. If I'm interpreting that correctly, then the behavior from 1% pressure to 0% pressure on the accelerator is incredibly disjoint and likely unsafe for an untrained driver.

What am I missing?

Volker.Berlin | 11 januari 2012

That's my interpretation of the quote. If I'm interpreting that correctly, then the behavior from 1% pressure to 0% pressure on the accelerator is incredibly disjoint and likely unsafe for an untrained driver. (brianman)

I agree, I had the same thoughts. It is not explicit in the quote, but after thinking about it my understanding is that easing up slowly will engage regen, and when you finally let go of the pedal, regen will remain at maximum level, even though you are in fact "off" the pedal. Only if you "ease off" the pedal in a single rapid move, the drivetrain will be disconnected and the car will coast.

Now, of course, you could argue that there is a discontinuity when you slightly press the accelerator again, because that would instantly engage max regen... Not necessarily. It's all in the software, don't forget that there is no physical connection between the pedal and its effect on the drivetrain. If implemented cleverly, smoothly pressing the pedal down could smoothly accelerate from coasting (if coasting before) or could smoothly decrease regen (if coming from max regen), all in the same pedal.

Similarly, for the complete effect, when pressing the brake pedal while coasting, smooth deceleration should be achieved by regen, and brakes should only be applied if you either heavily stomp on the pedal, or smoothly continue to press it further down beyond the point where regen is already maxed out.

Bottom line: The are many more ways to implement this than most of us can think of at first sight. I have been in the software business long enough to know that sometimes, simple implementations result in complex and counter-intuitive user interfaces, and simple and natural user interfaces sometimes require very, very complex implementational details. Just because the implementation sounds (or is) complicated doesn't mean the user (driver) must or should give any explicit thought about using it.

It's the ultimate goal to create the most efficient mix of acceleration, regeneration, coasting (and energy-wasting regular braking), without the driver knowing or thinking about it. It's only us techie early adopters who really want to know what regen is and when it engages. When have you last thought about which gear is currently engaged while driving an automatic transmission? If it works, you don't care one bit.

brianman | 11 januari 2012

Elaborating on previous post...

My understanding for the Roadster is that it's something like this bucketing for accelerator pressure:
1. 100%: power spent, high acceleration (+5)
2. 95%: power spent, mid acceleration (+2)
3. 90%: power spent, low acceleration (+1)
4. 85%: power spent, steady speed ( 0)
5. 80%: coast (-0.1)
6. 50%: low power regen, with slight deceleration (-0.2)
7. 25%: mid power regen, with light deceleration (-0.5)
8. 0%: high power regen, with low deceleration (-1)
9. 0%, and brake: no regen, mid to high deceleration (-2 to -5)

From the BMW quote, it sounds like:
1. 100%: power spent, high acceleration (+5)
2. 95%: power spent, mid acceleration (+2)
3. 90%: power spent, low acceleration (+1)
4. 85%: power spent, steady speed (+0)
5. 80%: coast (-0.1)
6. 50%: low power regen, with slight deceleration (-0.2)
7. 25%: mid power regen, with light deceleration (-0.5)
8. 0%: no deceleration ( 0)
9. 0%, and brake: no regen, mid to high deceleration (-2 to -5)

Bucket 8 on the BMW strikes me as troubling.

Transitioning through buckets 7-9...
Roadster: -0.5 to -1 to -2
BMW: -0.5 to 0 to -2

Disengaging the accelerator in phase 8 speeds you up?

brianman | 11 januari 2012

Correction:
"speeds you up" should read "decreases your deceleration rate"

Disengaging the accelerator from bucket 7 to 8 decreases your deceleration rate.

Volker.Berlin | 11 januari 2012

See my above post. You assume a fixed mapping between pedal travel and acceleration/regeneration. That's one way to implement it, and it has some appeal merely because it is so simple, but it's not necessarily the way it should be done.

brianman | 11 januari 2012

@Volker.Berlin
If you mean the "simple linearish numbers", that was just to simplify the description...or at least attempt to. I don't expect it to be linear; in fact, I expect the location of phase 4 to be speed sensitive and phase 5 to be sensitive to both speed and friction/gravity/etc.

Anyway it sounds like we're seeing it about the same.

IMO, either of these sounds predictable and understandable...
A. Accelerator does zero power (0% pressure) TO full power (100 pressure), no regenerative. Decelerator does zero regen (0% pressure) TO 100% regen (N% pressure) TO 100% regen with braking (100% pressure).
B. Accelerator does M% regen (0% pressure) TO 0% regen (N% pressure) TO 100% power. Decelerator does M% regen (0% pressure) TO 100% regen (L% pressure) to 100% regen with braking (100% pressure).

Where...
1. L, M and N are all in the range (0%, 100%), but not necessarily the equal to each other.
2. There are no discountinuities along the "TO" transitions.

I was under the impression that the Roadster does B; I don't know whether the Model S does A or B. BMW sounds like it's doing some scary C hybrid.

Volker.Berlin | 11 januari 2012

For all we know, we can safely assume that the Model S behaves as the Roadster, with the addition of a user-adjustable setting. It has been invented (well, implemented) by Tesla, Roadster drivers are consistently very happy with it, why should Tesla do anything different in the Model S?

brianman | 11 januari 2012

I'm fine with either A or B. The difference between them being that the coast point is a "passive driver state" for A whereas B places that at some non-0% pressure location of the accelerator.

B is likely statistically safer (no accelerator acts as a parking brake via regen), but I prefer A when driving long distances (somewhat like cruise control).

"why should Tesla do anything different in the Model S?"
I wasn't meaning to suggest they should, just that I haven't heard whether they've decided to tweak it from the Roadster or leave it alone.

Brian H | 11 januari 2012

I wonder if "a coasting mode" is something initialized first, and then the described behavior occurs; otherwise, normal regen.

Volker.Berlin | 11 januari 2012

Brian H, if regen is adjustable in the Model S, then setting the level to zero is effectively "initializing coasting mode" as you describe it.

Timo | 11 januari 2012

I just realized that Tesla implementation is not as intuitive as it sounds in engineering point of view. When you start to drive even slight pressure on the accelerator pedal produces torque, so in order to regen at all this pedal position to applied torque has to be variable and not constant.

IOW if you play this in other way around with fixed setting then you gain coasting with foot off the pedal and anything before that just slows down the deceleration caused by other losses if torque is not enough to maintain steady speed.

Quite a feat to make that feel natural. Electric engine does not follow quite same rules as ICE so "engine braking" has to be done artificially.

Volker.Berlin | 11 januari 2012

Timo, that's right. It's much more complex than it seems on the surface, but the nice thing is, that it does not *feel* complex. Discussing this kind of things is almost futile, because it is incredibly hard to imagine a simple behavior based on a complex description of the implementation. You'd need to actually test drive the implementations discussed here to be able to evaluate them. And that's what Tesla has done (lot's of user testing, I am sure) and what has resulted in the regen implementation known from the Roadster.

EdG | 11 januari 2012

Is it so difficult?

  • If you push the pedal down at an increasing angle X, the car applies torque, increasing till the pedal hits the metal and you're at max torque
  • decrease the pressure so the pedal rests at X(previous max angle)-N degrees (N between 1 and 5?) and you coast
  • let off more (N > 5) and you go into regen which increases until X is zero, at which point you're at maximum regen
  • increase X slightly and you're in coast mode
  • increase X more and you're back to applying torgue
  • braking is simply applying the brakes - independent pedal
  • Doesn't that cover it?

Timo | 11 januari 2012

Difficult thing is that that point two in your list has to be done manually and it can't be constant. If you use constant model then you don't gain regen ever, you coast when your foot is off the pedal.

Timo | 11 januari 2012

"done manually" -> "done artificially". Distraction.

EdG | 11 januari 2012

The magic of software. Only a very few things in the Model S are not allowed to be programmed. Like the brakes.

jbunn | 11 januari 2012

My preference (as if anyone asked) would be to have little to no regen on the accelerator pedal. If I need to shift my foot off for a moment, I don't want to feel like I'm hitting the brakes. What I'd prefer is something akin to the normal slowing engine compression causes when you take your foot off the gas pedal in an ice vehicle, down to just the air and rolling friction you get when you go into neutral on a manual tranny.

Coasting is inherently more efficient than regen (I'm sure this statement will generate some back and forth). While regen does convert motion into power, it does so at a loss (80% efficiency more or less). I can then use that juce to speed up again, but why not just coast? I don't have the regen loss, only the normal rolling and air losses.

Now I do want regen on the brake, and I want it to max out as much as possible before having to brake with friction (that's at total waste). Wonder how that handles, however since all regen will come from the rear wheels, and in conentional cars a disproportional amount of braking comes from the front wheels. Will braking function be better on a Tesla, especialy on curves and different traction surfaces? Looking forward to seeing...

brianman | 11 januari 2012

@jbunn - Yah, that's my A description. As previously noted, I'd prefer that as well.

Volker.Berlin | 12 januari 2012

I think all that should be said has been said in this and the other thread linked above. I'm trying hard not to repeat everything over and over... But here's one aspect that has not yet been mentioned with regard to its effect on regeneration:

Wonder how that handles, however since all regen will come from the rear wheels, and in conventional cars a disproportional amount of braking comes from the front wheels. (jbunn@hotmail.com)

The main reason for the disproportional amount of braking on the front wheels on ICE cars, as far as I know, is weight distribution. For one, usually the engine weight is on the front wheels, but what's even more important is a relatively height center of gravity that shifts most of the car's weight to the front axle under deceleration.

This is entirely different in the Model S, because the weight distribution between axles is 50:50, and this proportion does not change much under deceleration thanks to the extremely low COG. Therefore, I'd assume that even under deceleration, the rear axle should be responsible for almost 50% of the available traction (maybe even more, if different tires are used on the driven axle), which is a good thing to start with for regenerative braking (which inherently only affects the driven axle).

This of course only covers braking on the straight. One would hope that traction control is smart enough to release regen before the car suddenly oversteers when using regenerative braking in a curve. In fact, Roadster owners reported that the Roadster does precisely this, but it can be a bit hazardous because the driver must react quickly and put the foot on the brake pedal to engage all four friction brakes in this situation, otherwise the car would move on unbraked.

ggr | 12 januari 2012

I don't drive around constantly slowing and re-accelerating, unless there a traffic-based reason to do so. No one is stopping you coasting; your foot just sits in the right position to not provide current to the motor! I tried this last night, the hardest part was dividing my attention between the ammeter and the road. But I'm an instrument rated pilot, so even that wasn't terribly hard.

Anyway, I'm bowing out of this discussion now. I'm talking from my experience driving four different kinds of vehicles (Roadster with regen, automatic, manual including engine braking and coasting, and large trucks with valve lifters). Roadster-style regeneration IMNSHO is just wonderful.

EdG | 12 januari 2012

Nothing to complain about here? Time to move on to another topic until test-drive-time.

Nothing like reality killing a perfectly good topic of conversation. :)

JohanH | 12 januari 2012

I currently drive a Mitsubishi iMiev. Yeah, it’s not a Roadster or Model S (16 kWh battery, 0-60 in 13 seconds) but it’s an EV with regen. It has three modes for driving forward: D, B and C where D (drive) is normal, B (brake) is extra regen and C (coast) is less regen. I use D 99.9% of the time, B a few times when going down really steep inclines. I have never used, and I’m sure will never use C mode for the reasons described well in this and other threads by anyone who’s ever driven an EV (NO NEED!).

However, one thing comes to mind: My ICE (yes I have one of those as well) is a Volvo. It has a function called HDC (Hill descent control) which basically is just a sensor that senses when the car goes downwards really steeply and gears down more than normally when you let go of the gas pedal in order to get maximum engine braking (the ICE’s nearest equivalent of regen). I think it’d be cool to have feature like this in any EV. Maybe especially in the coming Model X which will be more “sports” and “off road” oriented?

mwu | 13 januari 2012

Not to try to take anything away from Tesla's engineering, but as a software engineer I don't think having a moving range of coasting before regen based on pedal position is very difficult. You simply keep tabs on the position of the pedal and use a (probably constant, but it also could learn) threshold or multiple thresholds to ensure that small variances keep the power regulation to the motor at the same rate. A let-off within a certain range is interpreted as coasting, and once you leave that range, you're in regen which should increase intensity as you let off more. Any time you reach regen while letting off or reach a faster speed while pushing the pedal further, the values the thresholds are based would update... All this would of course happen many times per second.

I would imagine the difficult parts are in controlling power rates / frequency to the motor and regen rates rather than when coasting should begin and when regen should begin. Not to mention how the battery pack charging and energy consumption has to be balanced out.

Another way of saying it: I think interpreting what to do based on pedal positioning is probably fairly easy, but going from what it needs to do to determining what fine adjustments are necessary to carry it out isn't.

Douglas3 | 14 januari 2012

I'll say it again.... if you haven't driven a car with regen on the accelerator, don't knock it. You'll be surprised. I greatly prefer the driving experience with it.

Brian H | 14 januari 2012

Since the X will have 4WD, that changes the calculus, as it means 4WRegen, too. I assume there are some (not drastic) differences in COG and such that will affect how it drives. Interesting comparisons coming soon!

Volker.Berlin | 19 januari 2012

Single-pedal driving: easy

Like the Tesla [Roadster], the [BMW] ActiveE is easy to drive on a single pedal. The regenerative braking is relatively aggressive, but very well modulated, and it takes only a few minutes to learn just when to lift off so the car comes to rest just behind the car in front at a stoplight.

If the needle in the energy meter is exactly vertical in the "Ready" position while underway, the car is gliding, using an accelerator position drivers learn to find when they want momentum to carry the car as far as possible without the slowing effects of regeneration.
http://www.greencarreports.com/news/1071880_bmw-activee-electric-car-fir...

Robert.Boston | 19 januari 2012

If the needle in the energy meter is exactly vertical in the "Ready" position while underway, the car is gliding
Having that visual feedback in the Model S would be very helpful, at least until we learn to feel it intuitively.

dborn @nsw.au | 19 januari 2012

Robert, have you driven the roadster? I have, and 10 minutes is about all it takes to get the "feel". It is amazingly intuitive.

Douglas3 | 19 januari 2012

+1 dborn. In my experience, only people who've never tried it complain about regen on the accelerator pedal. Those who have love it.

Also, not that you need it or anything, but there's a nice big power meter in the Roadster's instrument cluster. When it reads zero you're coasting.

William13 | 19 januari 2012

At the factory event I was told that there would be a small amount of creep. Now that the roads are icy in South Bend I am hoping there is very little or no creep. I remember my car accelerating, while pushing on the brakes and the front wheels locking. I had to put the car into neutral to get it to stop.

gjunky | 19 januari 2012

I am not trying to argue if regen is better or intuitive but allowing true costing (no torque applied by the engine, no regen) is always more efficient than regen because regen will never capture all the energy back. None of this will make any difference once you come to a stop at which point both will be the same.
(You would have come to a stop earlier with regen but with more power in the battery pack vs. driving further with less energy left.) To travel the same distance, starting from a constant speed, letting the car coast will be more efficient. Not always practical in real life of course. This doesn't work for stop signs (you are coming to a stop) but coasting would help when you can anticipate you have to slow down slowly for traffic ahead but don't need the regen to slow you down (more than the rolling resistance and wind resistance)

VincentA | 19 januari 2012

I heard from a reliable source that the Model S alphas and maybe the betas had regen adjustable by the driver. Whether they leave that in the production vehicle is another matter, of course.

EdG | 19 januari 2012

While you're correct that maintaining as close to constant speed as possible will be the most efficient, I have every intention of doing almost the opposite when I drive.

  • Given the fun of jumping out to a quick start,
  • knowing I'll get much of the energy back with regen, and
  • understanding that the value of the total energy lost is negligible with respect to what I usually do (even if I try to save as much gas as possible) in an ICE car,

I'm hoping to have more fun than ever before while knowing I'm wasting almost nothing.

Let all the other drivers think I'm wasting gas.

Brian H | 19 januari 2012

EdG;
Yeah, it will be real hard for some to get over that "hypermiling" mentality in normal driving. However, on borderline range-runs it will come back full force!

Volker.Berlin | 20 januari 2012

VincentA, we can legitimately assume that regen will be user-adjustable in the Model S:

By luck, one of the first I introduce myself to is Drew Baglino, who happens to be directly involved in setting up the Model S regen braking system. [...] When I pleaded for driver-adjustable regen on the Model S, he said this was under consideration. (Two days after the event, while writing up this report, on a whim I e-mailed Tesla boss Elon Musk and repeated my plea for driver-adjustable regen. Three hours later, I got this reply: "I totally agree that regen should be driver-adjustable and it will be on Model S." The message was Cced to JB Straubel, Tesla CTO.)
http://www.thetruthaboutcars.com/2011/10/tesla-model-s-customer-blog-bet...

Teslamodels4me | 25 januari 2012

Yes, ask and you shall receive. : )

Pages