Hacker News Reader

Comment by gen220 | Hacker News Reader

You should have higher expectations of your vehicle.

My 2016 ICE car's "miles left" meter is accurate to +/- 2 miles from the moment I top up the tank (80% highway driving, 20% hilly city and rolling country roads).

IMO, accurately telling the vehicle operator how many miles of juice you have left is a KPI, as it informs when you'll need to plan for refueling.

Having driven an EV for a few weeks in identical conditions, this inaccuracy is probably the major contributor to "range anxiety". I have no idea whether I'll need to recharge in 60 miles or in 25 miles, and that's totally unacceptable in most parts of the US (where there aren't available chargers every 5 miles of your trip).

View on HN

> My 2016 ICE car's "miles left" meter is accurate to +/- 2 miles from the moment I top up the tank (80% highway driving, 20% hilly city and rolling country roads).

No way. That’d be accurate to under 1% - well within the variation you’ll get just from different air densities (temperature, pressure, humidity). Forget about A/C on or off, tire pressure, etc.

It’s gotta be tricking you somehow.

We regularly do road trips of 200+ miles, and we get 350 miles on a full tank of gas. At the beginning of the trip, the GPS will say something like "250 miles to your destination" and the car will say "180 miles left"

I regularly pull off the highway to fill up with around "20 miles" left, when there are ~90 miles left in the trip (e.g. it was dead-on accurate for the last 160 miles), according to the GPS. I agree that it's spooky.

I think, when making estimates for "miles left", our car must be doing some calculation like (rolling estimate of mpg) * (gallons left).

It's also possible that it keeps a buffer of ~1 gallon of gas that it never tells us about, when quoting the total range, and it eats into this buffer without telling us. There's certainly a bunch of tricks they could be pulling off.

The point is, from a driver's point of view, I have a trustworthy number. If they need to pull off some tricks to achieve that trustworthiness (e.g. low-balling total range), that's OK with me.

With EVs, "total range" is a marketing term, so they have a strong incentives to fib about it. Unfortunately, this makes the vehicles worse to operate.

“It's also possible that it keeps a buffer of ~1 gallon of gas that it never tells us about, when quoting the total range, and it eats into this buffer without telling us.”

I am fairly confident that something similar is happening with my Honda. I thought I was going to run it dry, but then it just sat at 10 miles remaining for at least another 15-20 miles before I reached a station.

Most Honda models have a ridiculous reserve, something like 3.5 gallons. I still regret informing my mother of this fact...she never ran out of gas until I did.

> It's also possible that it keeps a buffer of ~1 gallon of gas that it never tells us about, when quoting the total range, and it eats into this buffer without telling us.

This is quite likely. My car supposedly has an 11-gallon gas tank, and if I refuel right when I hit 0 miles remaining then it only takes 10 gallons of fuel to fill the tank.

Fun thing about a car: They have computers that measure all those factors! Your engine knows the air density, air temperature, the rough gravity vector (used for things like hill start assist), lots of info on tire performance, speed, fuel flow, etc etc etc, with more modern cars getting things like GPS details. The estimate they use for range is based on the average fuel economy you get, not some fake model made by the EPA that doesn't represent reality.

This has been true in Toyotas since at least 2004, so I would hope it could be true for a "Luxury" car from a "tech" company.

It doesn’t measure those factors in advance.

It doesn't have to. It measures them in real time, and can integrate over time.

Astronomers also don't see into the future but some basic calculations can let them make very accurate predictions about where the planets will be in a month.

What the heck is your point?

Okay, so if I give you all the data from my car, and say it has 10.3 gallons of gas, you’re telling me you can predict how many miles it will go to within a percent?

No, you can’t. Because I might go uphill or downhill. It might rain. It might be hot or cold.

The car is doing something tricky (aka lying) as discussed above.

I think if you gave us a hidden buffer of 1 gallon (i.e. we actually have 11.3 gallons, but I report that you have 10.3 gallons, and selectively burn some of that buffer without reporting it), and you drive in conditions that fit within the footprint of your prior 3000 miles of driving history, yeah we probably could guess within a percent.

To the best of my understanding, this is what most modern ICE cars do when they give you the "miles remaining" metric. I don't know how else they could be so outwardly accurate.

This falls under the category of "trick", but it's actually a trick that helps me by allowing me to plan my trips better.

Giving the "current estimate of actual miles remaining", as a multi-variate function of all the things you mention, is actually less helpful and more stress-inducing – even while it's technically providing more information, it's information that isn't very helpful.

A Tesla has exactly what you describe in the Energy app, and it is very accurate, always within 1-2% in my experience. The newest version also has a breakdown of which systems are drawing power.

But that’s off topic; this article is about the car not meeting its EPA range, because the EPA test is very synthetic[1] and doesn’t match most people’s driving habits. I find InsideEVs’ “70 mph range test”[2] series to be far closer to my personal experience.

[1] https://www.epa.gov/greenvehicles/testing-national-vehicle-a...

[2] https://cdn.motor1.com/images/custom/bev-ievs-70mph-range-te...

Maybe it gives the user an underestimate then wastes gas later to meet it? :D

The difficulty is because of the difference in operating regimes of the power trains.

An old ICE has to run within an RPM range, so the top end on MPG is set below infinite, and the bottom end is limited by the time to ramp the engine to its max consumption.

Electric vehicles have neither of these issues. On a downhill run that is long enough you could in theory fully recharge your battery, and you can always send full power (and resultant fuel consumption) to the engine.

On top of this EVs are operating at much higher efficiencies so the effective size of the tank is smaller.

This is not to say range estimation couldn’t be improved, only that it is a much more challenging model with EVs.

>On a downhill run that is long enough you could in theory fully recharge your battery

Such a road would by identifiable by the car's GPS.

> this inaccuracy is probably the major contributor to "range anxiety"

Btw, this "range anxiety" term, is it designed to make people seem crazy for wanting a long-range car?

That's a fun idea :)

It refers to the anxiety people have about not having enough "range" left in their car to reach their target destination. The symptom is looking at your "% battery remaining" every 5 seconds, and cursing your car when it drops "3 miles" of range in 1 mile of idealized highway travel. Unfortunately, it's a well-founded anxiety.

You're right, by characterizing the problem with a focus on the person (with anxiety) instead of the (misleading) vehicle, it serves the manufacturer more than the consumer.

Hmm, seems my suspicion about the origin of the term was right:

> The term "range anxiety" was first reported in the press on September 1, 1997, in the San Diego Business Journal by Richard Acello referring to worries of GM EV1 electric car drivers. On July 6, 2010, General Motors filed to trademark the term, stating it was for the purpose of "promoting public awareness of electric vehicle capabilities"

Can't think of any other product where the user gets blamed like this. I really do have "MPG anxiety" in my ICE car, always checking the gauge, but nobody calls it that.

Reply on news.ycombinator.com