Indicated range: need an option for user calibration of miles / kWh

Indicated range: need an option for user calibration of miles / kWh

Apologies if this topic has been covered previously - if so, please link to the relevant posts.

I need to set the mileage gauge (standard "miles" of range left) to more accurately reflect my driving pattern.

I am a 2014 P85+ owner and I've found that the "miles range" gauge is consistently off by 30% in the wrong direction for typical 75 MPH California highway driving. In other words, an indicated range of 245 "miles" equates to a real highway range of 170 miles for this type of efficient highway cruising in this car.

This issue is about having a useful range indicator. It isn't about how efficiently I'm driving the car, or how the interface is designed for simplicity, or that the indicated miles are not meant to reflect actual driving and that I can review the energy meter, etc.

I think the most straightforward way to deal with this is:

- enable the input of a user calibration of miles / kWh that is used to generate the indicated "miles" of range, perhaps using a different color from the Tesla calibration (does anyone really drive 55 MPH anymore?)
- enable an optional display of kWh and observed miles / kWh on the display

ye | 10 mars 2014

Makes sense to me.

I guess, in the meantime, until Tesla implements this, if they ever do, the closest thing to what you want is the projected range that the energy meter shows on the center console. But maybe you already know this, because you mentioned reviewing the energy meter.

mgboyes | 10 mars 2014

Pretty much every car I've owned has had a "range" readout on it. They have all worked in the same way: by estimating my actual range based on my fuel consumption in the recent past (last hour, last 50 miles, whatever).

I cannot fathom why Tesla thought it was OK to put a different number in the middle of the power gauge in the center of the instrument cluster like they have. At the very least they need to add a third option (currently you can have "rated" or "typical", when I suspect what everyone really wants is "estimated actual").

mgboyes | 10 mars 2014

Also, as a P.S. to the above - why would you ever design a user interface that is potentially so hugely misleading, especially one which to some extent is giving safety related information?

If someone gets into a Model S, looks down, sees "265" in big friendly letters underneath a happy green charge bar, and then runs out of juice 200 miles later then IMO the fault is with the user interface, not the user. Electric cars need to be at least as easy to use as ICE, and this is one area where the Model S really is not.

J.T. | 10 mars 2014

@mgboyes You seem to be saying that during the 200 miles it took to run out of juice the big friendly letters (numbers) undreneath the happy green bar became invisible.

mgboyes | 10 mars 2014

@jtodtman I realise I am using a slightly exaggerated example, but I am trying to think about a Model S user who is not an early adopter tech-savvy person. I am trying to imagine lending the car to my father, who might find all the gauges and dials and newfangledness a bit odd.

The point I am making is that the range gauge is given pride of place right in the middle of the instrument cluster and it basically always lies. You might as well have a speedo that tells you what speed you would be doing if your wheels had a different radius, or a satnav which gave you an ETA as if you were in a helicopter. Sure, you can do some rough math in your head to convert these things, but why should you have to be made to think while you're already busy piloting 2 tons of aluminium at 25 meters per second in the driving rain?

Spammy | 10 mars 2014

BTW - I'm pointing Tesla personnel to this posting, please continue to post your thoughts so that I can get some Product Management attention on this...

jordanrichard | 10 mars 2014

Don't have my car yet, but isn't the "projected range" based on recent driving, displayed on the center screen, on the energy screen?

mgboyes | 10 mars 2014

@jordanrichard yes it is - I think the only thing missing is the ability to have that same number displayed in the middle of the instrument cluster.

Spammy | 10 mars 2014

The energy screen's projected range is a moving average computed over two different mile windows. In practice, their particular choice of moving average is too variable for use as a typical "miles remaining" range indicator IMO.

What is currently displayed in the center of the instrument cluster is "Rated Range". I believe they should enable a change of default display there to "Remaining Range", computed using a long window moving average, perhaps with a smoothing function added, much like other car manufacturers do now. I also think they should augment this with a small indication of "kWh / mi" consumption, much like other car manufacturers are indicating fuel economy.

J.T. | 10 mars 2014

@mgboyes I understand your concerns vis a vis a driver unfamiliar with the vagaries of range in an electric car, but I'm not sure there can ever be an adequate solution.
If it isn't driving style, it'll be outside temperature. If it's not that it'll be elevation. All of these factors can take a tremendous toll on range.
It's a thinking person's mode of transportation. You just can't take it for a spin without a plan. Any owner who lends someone their car for an extended time has to make this clear.

mgboyes | 10 mars 2014

@spammy +1 for the most accurate prediction algorithm they can come up with - and I bet they could do something very clever indeed, especially if you start factoring satnav data into the mix (often the car knows where you are going, and on what roads).

But to start with, +10000 for anything that's based on actual data rather than a fixed hypothetical model.

mgboyes | 10 mars 2014

@jtodtman I am pretty sure Elon's vision is for it to be everyone's mode of transportation!

A gauge extrapolated from recent actual usage will automatically factor in elevation, outside temperature, driver style, weight of passengers, just like it does in every ICE car.

Spammy | 10 mars 2014

@mgboyes: sure, good idea - perhaps you can augment the consumption calculation using topographical information, along with using a few other strongly correlated efficiency measures like speed variability (stop and go) or just current speed, but I think it's a UI matter that needs some study and experimentation to find the most agreeable combination of factors that drive the UI.

That said - until they get a super cool and useful version, they should just do the basic moving average on the center display because this is a confidence damaging oversight IMO.

J.T. | 10 mars 2014

@mgboyes Have you ever actually counted on the "Miles to Go" readout in your ICE car? With a gas station on every corner there's no reason to. When the Superchargers are on every highway and there are adequate High Amperage chargers in every city Elon's plan will come to fruition, but until then anyone who plans a road trip without doing some serious calculating is simply foolhearty.

jordanrichard | 10 mars 2014

It would be nice if it were known how many miles driven are used to calculate the "projected range". If one just came off the highway and is now tooling around town, the projected range is going to be seriously skewed. If one knew that they were taking into account the last say 10 miles, and one obviously knows what those last 10 miles of road were, they could get a better sense of the accuracy of the "projected range"

I don't see what purpose or benefit a kWh/mi indication gives you. Most people are not electricians, so throwing kWh/mi numbers at them isn't going to help. No one can relate to it.

My Volvo, like many other cars tells you how many more miles it has to go before being empty. This is not based on the EPA mpg rating, it's based on what the car has done for the last "X" number of miles. People don't care how the car figures the "miles to empty" number, they just want to know how many miles.

So why doesn't Tesla put that as the only range number to be read?

Spammy | 10 mars 2014

@jtodtman: I have used my "miles to go" indicator in my ICE and hybrid cars and they're pretty good - maybe +/- 5%.

I see no reason to suffer with a consistently 30% wrong large display, especially when it's so easily fixed.

There may be some reluctance on the part of the company to display a large number with a substantially reduced predicted range from a PR perspective, but as someone pointed out, this is a safety matter in addition to a usability one.

redacted | 10 mars 2014

I have one of those. I (for example) multiply the rated range by .72 and presto!

The other way to do this is to bring up the energy screen, set "30 miles" and set it to average. Presto!

Spammy | 10 mars 2014

@redacted: So, do you do this for other readouts you use regularly? Say, your instant read cooking thermometer reads 180, but I multiply by 0.7 to get the real temperature "presto!". My measuring cup says 3/4 cup, but I multiply by 0.7 to get the real measure, etc

Not particularly practical, and I doubt you would be happy if you'd spent double or triple the normal going price for each of these tools and had to apply a 0.7 multiplier on their basic readouts...

mgboyes | 10 mars 2014

@jtodtman I take your point that today the lack of superchargers every 30 miles means that on long journeys a Tesla owner has a different duty of care about planning their progress. (I'm actually offering out a weekend loan of my car this summer as a prize in a charity auction, so have already started thinking about the level of education I will have to give the winner before I let them take the car away!)

So I can also see that Tesla might have decided to subtly encourage that sort of planning through the way the car interacts with the user, the messages it gives, the language used etc.

So why would they ever want the car to imply to a user that they have more range than they really have? For most people in most situations rated range is an over-estimate. Why show it so prominently when really it should only be mentioned in the sales brochure?

redacted | 10 mars 2014

@spammy actually, yeah. My wife's car's clock, for example, was consistently set 7 minutes ahead. My outdoor thermometer reads four degrees too warm when below 0F. Translations are good practice for the brain. And .7 of a 3/4 cup measuring cup would be 1/2 cup, that's one I'd only have to do once.

I think there might be more useful ways for Tesla to do this, and expect it will. I theorize though that more and more information actually creates range and mileage anxiety that such a car shouldn't have. I suspect that people would enjoy their cars more if Tesla took it all away and gave you a meter like a normal car that only showed approximately how far you were from E.

ye | 10 mars 2014

spammy2013-1 said, "The energy screen's projected range is a moving average computed over two different mile windows. In practice, their particular choice of moving average is too variable for use as a typical "miles remaining" range indicator IMO."

I don't have the car, but based on what I've read, I think it works like this: The energy graph can be set to show the last 5, 15, or 30 miles. The projected range is based either on that number of miles, if set to "average", or on the last tenth of a mile, if set to "instantaneous".

Thirty miles seems like enough to get a reasonably stable number. Do you find that the projected range is still too variable when it's set to "average" and the graph shows the last 30 miles?

J.T. | 10 mars 2014

I think spammy hit on the head. There may be some reluctance on the part of the company to display a large number with a substantially reduced predicted range from a PR perspective

It's all about test drives. Every time I give a test drive people ask about that number and then the range. I'm sure it's the same with Showroom test drives. I suppose until they come up with an algorithm that works under all situations perhaps instead of the big numbers they should just have IT DEPENDS.

jordanrichard | 10 mars 2014

I think the issue is, that number on the energy screen should be the only number given and should be the number that is under the speedometer. Why does one ned to know how many kWh/mi one is using? If one sees that there range to empty is low for where one is going, slow down and the that range number will go up.

Generally speaking , only electrical engineers want to know about the kWh/mi efficiency of the motor. Drivers want to know how far can they can go before needing a charge.

It's like RPM gauges. Unless you drive a stick shift, why do you need to know how fast the egnine is spinning via a RPM gauge?

mgboyes | 10 mars 2014

@redacted I think you would be right that having a just a green/orange/red bar and no numbers might suit some people better... *if* we were at a point where there were superchargers every 30 miles and where you'd just drive to the nearest "gas" station whenever the gauge went red.

@jtodtman I have test driven AMG Mercedes and Ferraris and it never bothered me that at the end of my "spirited" ride the miles remaining number was hilariously small. That's what happens in a performance car when you drive fast! A test drive is not a realistic environment in which to test fuel economy in any car. It's not like I didn't notice that on my 10 mile Model S test drive the car lost about 30 rated miles!

All we're asking is for a third option to be added to the current two, controlling what is shown in the center of the power meter, so for test drives Tesla can keep setting it to rated anyway.

Bighorn | 10 mars 2014

I think there's merit to making the 30 mile average projected range available on the dash. I just run it on half of the 17" screen when I'm making strategic decisions about driving economy. Comparing that to miles remaining on the GPS is imperative--again a function showing on the 17" screen.

@jordan Wh/m will become second-hand to you in no time.

J.T. | 10 mars 2014

@ mgboyes My test drive comment was not about how much energy was used but more about that number being a sales tool., like spammy suggested.

Spammy | 10 mars 2014

@ye: yes, 30 miles appears to be an insufficient integration window for stable results. There are straightforward ways to do long term running averages of time series data like this that have been used by other car manufacturers to produce very usable and stable estimates.

In my case, I'd expect that running estimate to deliver a predicted (miles / kWh) that is about 27-32% lower than what is used to produce the "rated range" used for the center displays. As a consequence, the large number in the center will probably read "170 miles" instead of "245", which will cause the Tesla marketing people to frown. This should be mitigated by making it an optional display parameter IMO.

The current situation is poor - there is a very large number displayed on the center display cluster that is clearly intended to be used as the primary gauge of "full / empty" and it's completely, always wrong. It's a dumb thing for a smart company to do IMO.

NKYTA | 10 mars 2014

"and it's completely, always wrong"

Actually it is not - many MS drivers consistently "beat" that number. Please don't be so strident.

We've heard that Elon wants to make the Nav+Energy app "much better", which might include what you are suggesting. I, for one, am looking forward to that.

mgboyes | 10 mars 2014

@NKYTA if your driving style means that you exceed the rated range, then it's just as wrong as if you fall short of it. Of course technically there might be some person out there whose driving style means that every journey they ever do exactly fits the EPA mix of terrains and speeds so that they always end up exactly hitting 265, but it's not likely is it?

Rated range is like a stopped clock - just because every once in a while (twice a day in the case of the clock) it gives the right answer doesn't make it any less useless.

jbunn | 10 mars 2014

It's pretty likely that it gives many of us the right answer. I find my range estimate amazingly accurate. I leave a 20 mile buffer when traveling, and that works fine. On my last road trip, the miles driven and (starting range - remaining range) were about 1% apart. Typically, my error is within a couple percent either way.

(and I live in California, where 70 will get you rear ended by a granny)

(and I like driving a performance car)

NKYTA | 10 mars 2014

@mgboyes, well there you have confirmed it. If it is right "every once in a while" then clearly it can't "always" be wrong. Never, ever, use always and never (to the OP).

"always end up exactly hitting 265, but it's not likely is it"
I've done it.

We've been talking about how the Energy App could be better for more than a year, and we've heard rumors that Tesla is working on re-vamping it. They haven't even said it is "coming soon" yet, but they have a lot of stuff going on and this is pretty minor for me, who has been driving the car for 14 months. And it is on the longer trips that I get near rated, never around town - but around town I barely look at the rated range.

You can always go submit/vote for this on the Prioritized Software Enhancement thread if you are an owner/reservation holder.

mgboyes | 10 mars 2014

@NKYTA ROTFL at you trying to ban a word but using the banned word while you do so. However I'm not going to be dragged into a semantic argument with you as that would rather miss the point.

The fact remains that on average across all Model S drivers, and all their many different journeys, an estimated remaining range gauge based on some sort of historical sample of data will be more accurate, and therefore more useful, than a figure which is just "percentage charge times 2.65". Plus there's the additional benefit that it's what everyone is used to from ICE.

Brian H | 10 mars 2014

second nature, too. Maybe just as handy, firstly or secondly. <6-p

NKYTA | 10 mars 2014

@mgboyes, I was being ironical, clearly.

I'm not arguing that Tesla shouldn't make an improvement here or in the Energy app. I'm just stating my opinion that after you've driven the car for a while, this really isn't a big issue, at all.

When we get all the Nav/Energy improvements that everyone is asking

So go to the PSEL and vote for this idea if you choose.

39. s2. is much more important to me.

37. j. Display both energy & miles added on Charging Status screen, 161 (R&B, 10/28/12)
71. c6. Include projected range on instrument panel energy display, 59 (brookbot, 3/7/13)
114. n6. Option to enter user-defined distance for graph & data in Energy app, 20 (Jamon, 3/16/13)
k2. Display kW/Mile of a Rated Mile on the Energy screen – v4.0 (1.17.50)
4. p1. Ability to save nav destination as a favorite and to name the saved location, 622 (Portia, 10/30/12 & iholtzman, 12/25/12)
7. w1. With traffic & nav selected, adjust route time and make traffic indications visible, 534 (R&B and JasonK, 11/1/12)
9. v1. Ability to select alternate nav routes or to alter current nav route, 488 (R&B, 11/1/12)
34. h1. Nav system acceptance of location from an external source such as Google’s “send to car”, 178 (Electric Machete, 10/30/12 & paul@cable, 4/30/13)
39. s2. Forecast Rated Range remaining at nav destination taking terrain, weight, climate control, temperature, and other variables into consideration, 156 (MB3, 11/22/12)
43. u1. Ability to drop a location on the nav map, 141 (R&B, 11/1/12)
60. p5. Ability to display Nav map on instrument panel when not navigating, 82 (DouglasR, 2/9/13)
66. d5. Option to hide/unhide turn-by-turn directions box during navigation, 69 (bbmertz, 2/7/13)
77. t3. Restore prior display to left inst panel at conclusion of navigation, 50 (Dogface, 12/19/12)
87. h2. Selectable nav voice fade to driver only, 38 (DavidG, 11/4/12)
119. y8. Icon to toggle on/off the navigation path (to see traffic colors), 20 (WSE51, 2/12/14)
128. s6. Provide “are you sure” message in response to cancel nav selection, 17 (jukevargo, 4/8/13)

Bighorn | 10 mars 2014

Right after I hit "submit", I had second thoughts.

Spammy | 10 mars 2014

@NKYTA: The PSEL sounds like the right mechanism for this - IMO it should also be augmented with the opinions from this kind of thread - both in number and content.

WRT this not being a big deal for most people - I disagree. I think it's wildly inaccurate, especially for owners of the P85+ with 21" wheels, which has a systematic and predictable reduction in the (mi / kWh) constant that should be used for the gauge and yet, somehow, isn't reflected in the gauge.

This is an important modification that is needed to provide basic, expected car functionality to the Tesla S. Cars have fuel gauges that are accurate within +/- 5%, so is this a car or a science lab project?

I think most of the people reading and responding to this thread are capable of trip planning for the car as it is, factoring in all of the bugs and known deficiencies and calibrations required. Some of us are also capable of planning the orbital trajectories necessary for travel to Mars. I argue that most owners of cars expect not to have to carry around constants like "0.72" that they must apply to the most basic of gauges in order to do simple everyday driving.

J.T. | 10 mars 2014

@spammy I agree with most of what you say but your last sentence is incongruous. No one needs to do any calculations at all to do simple everyday driving.

donaldmeacham1 | 10 mars 2014

I too had not had any difficulty with the current system. I always give myself a 30 mile buffer and nearly always end up a trip with little discrepancy between my starting rated range and the remaining miles at the end of the trip. This is with city and freeway driving including 65-75 mph.

jbunn | 10 mars 2014

Just got off the phone with Tesla. The engineer who wrote the algorithm has announced his intent to commit Seppuku - the Japanese art of self-disembowelment to atone for his shame.

Hopefully, this helps alleviate some of the bad feelings over this gross inconvenience.

tes-s | 10 mars 2014

Would you want the "main" range readout to be the estimate from the energy screen average over last 30 minutes? Since last "fill up"? Or, should it be based on a calculation of kWh remaining and an enterable wh/mi? Should it include the "reserve" or not? Should you be able to enter your own reserve?

Spammy | 10 mars 2014

I've placed a feature request here:

Please vote for the feature and/or comment on it!

grega | 10 mars 2014

@spammy, you say you're doing typical highway driving, 75mph, and call it efficient highway cruising. But 75MPH actually has a substantial reduction in efficiency.

Does the graph here fit what you're seeing:

Also how to calculate range raises lots of interesting questions. For me, I'd rarely drive more than 100 miles a day, and when I do it'll be because I'm driving a long distance at higher speed. So theoretically I don't want my daily driving to calculate my long distance driving range. I don't want the first 15 minutes of my driving to be used to calculate the next 2 hours usage either for the same reason (time to get to the highway!)

The only sure thing is a graph of kWh remaining, which for most people would be meaningless.
Even "range 215 miles at 45mph" will differ depending on the driver.

Spammy | 10 mars 2014

@grega: yes, of course, the range reduction should track the power consumption effect of wind and rolling resistance with increasing speed. Note that the aerodynamic forces of form drag follow a cubic relationship between power and speed - so that (Power = Cd * (1/2 rho * V^3)) where rho is the air density and V is the car's speed. That's a big hill to climb - as V increases, power consumed increases by the cube. The Model S's drag coefficient (Cd) is very low, but it doesn't matter.

This and other things (like size of wheel) make a big difference in observed range - therefore a fixed coefficient of efficiency (miles / kWh) is incapable of providing a useful gauge.

WRT the alternatives and how accurate they can be - I think you can provide a long term rolling metric that will capture the essence of your driving and vehicle configuration. It certainly won't be worse than the current fixed coefficient.

Spammy | 10 mars 2014

BTW - re-reading the posting, the OP says "aero resistance increases with the square of the speed" which is correct, but to calculate power, you multiply force again by the speed, which leads to the cubic relationship between power consumption and speed.

Brian H | 10 mars 2014

The linked graph says about 340 wh/mi at 75 mph for an S85.

ye | 10 mars 2014

Power to overcome air resistance does go as the cube of the speed, but usually you're not as interested in that as you are in the force of air resistance, which goes as the square of the speed. The reason is that usually you have a certain distance that you want to cover, not a certain time for which you want to drive. If you drive faster over a fixed distance, you get where you're going in less time, so that partially offsets the increased power usage, bringing the total energy consumption down to the square of the speed rather than the cube.

ye | 11 mars 2014

spammy2013-1 said: "yes, 30 miles appears to be an insufficient integration window for stable results."

That's surprising to me. I suppose it might be true, but I can think of a couple of other possibilities that might explain the instability you see.I remember reading some time ago that the energy app switches back to instantaneous mode every time the car is restarted. Maybe that's what's happening?I might be mistaken about how the projected range is calculated in average mode. I don't remember how I came to the conclusion that it averages over the number of miles displayed by the graph. I looked at the manual just now and I can't find it there. Can someone with the car verify that, in average mode, the projected range changes when switching the graph from 5 to 15 to 30 miles?

spammy2013-1 further said: "There are straightforward ways to do long term running averages of time series data like this that have been used by other car manufacturers to produce very usable and stable estimates."

Do you know how other cars produce their estimates of the number of miles remaining, or are you just saying that there must be some way of getting an accurate and stable estimate because, in your experience, other cars manage to do it?

I am not sure what the best approach is. On the one hand, you don't want the estimate to exhibit large, meaningless fluctuations, but on the other hand, you do want it to reflect your current driving style. If you see that you're going to run out of energy before you reach your destination, and so you slow down in order to reduce your energy consumption, you don't want to have to wait 30 miles or more to see whether you've slowed down enough.

triss1 | 11 mars 2014

Although a few more options on the energy app might be useful, for my two cents, the current dash display against a fixed benchmark is useful. Keep in mind that the car already includes a gas gauge-like display--the large green bar that turns gold and then red as you run out of charge. The problem with using any measure of recent usage as the benchmark is that recent usage can change dramatically, especially in cold weather with the battery heater changing from off to on or on to off.

Having a fixed benchmark (300 kW/mile) tells you how much you need to slow down to hit the rated range and what you will accomplish if you do. Having the Rated line on the energy graph would not be nearly as helpful if the mileage remaining display was constantly changing.

Spammy | 11 mars 2014

@ye: For the purpose of this discussion, power (say kW) is important. The amount of power consumed by the car, which measures energy consumed per time (Joules / second or kW-hour / hour) indicates the degree to which the battery is being drained. So, power consumption depends on the speed cubed.

The reason this might be useful for a feature of this kind is that you could develop a simple physical model of energy dissipation that includes the aero and rolling friction effects of the car, combine it with the energy dissipation characteristics of the battery and motor that can then be used to "curve fit" the observed driving pattern and yield a more accurate prediction of driving range.

@triss1: Having an efficiency target is useful, yes, but not at the expense of having a real "fuel gauge." I think you can easily have both, but if I had to choose only one, I would choose to have a useful and accurate fuel gauge first. I also would choose to put it prominently on the center instrument console. This choice is consistent with every modern car. I have owned a Prius and a Hybrid BMW and they both have efficiency targets integrated into their instruments, but they also have prominent and accurate fuel gauges and "miles of range left" indicators that aid driving. The fact that Tesla is using a "range target" as the only fuel gauge is an aberration compared to industry practice - a choice that diminishes the utility of the Model S in comparison with the competition IMO.

J.T. | 11 mars 2014

@spammy The fact that Tesla is using a "range target" as the only fuel gauge is an aberration compared to industry practice
Isn't there a battery level indicator on your speedometer? Isn't that the "fuel gauge" you're looking for?