Battery gauge inaccuracy

[dmen]

I took a 70 mile drive today. I left the energy screen on display to track the percent battery charge remaining, and used a fresh enhanced trip recorder to track kWh used.
At 90% remaining, I used 1.5 kWh. (Blue bar looked about 1/10 below full)
At 75% remaining, I used 4.1 kWh.
At 50% remaining, I used 9.4 kWh. (Blue bar at top of "mi" label)
At 25% remaining, I used 14.7 kWh.
I plugged in at 19% remaining, at which point I used 16 kWh.

So total usable capacity according to those numbers apparently went up from 15 to 16.4 to 18.8 to 19.4 to 19.7 kWh as I drove. I didn't go to empty, but not long ago I determined the available capacity (full to display of 0 mi range warning) was 19.8 kWh. I hope it's still about the same.

Why did Ford bother adding a percentage display if it isn't accurate? Especially when taking a trip that will approach range limit, it does no one any good to see a falsely dropping percentage over the early miles.

 

[WattsUp]

As you drive, there are two sources of energy; the initial charge stored in the battery, and energy captured from regeneration. Do we think the trip recorder's "kWh used" figure includes both sources? (It necessarily would if it simply measured the energy consumed by the motor.) If it does, that might throw your calculations off a bit.

 

[dmen]

Yeah, it does account for recaptured energy. When slowing from highway speed to a stop, it sometimes goes down. And the percentile display sometimes goes up during those same decels. So recaptured energy is reflected in both the kWh used, and the state of charge.

And anyway even if it didn't, you wouldn't see the behavior I do on these long drives... The calculated capacity would get lower over the drive, rather than higher, as cumulative regenerated energy failed to be accounted for. But again, it does apparently account for this.

I think your questioning comes down to whether the kWh used is accurate as my interpretation assumes. I think it has to be accurate, because it is directly related to the displayed efficiency. When I have driven 10 miles, if my efficiency on that trip is say 230 Wh/mi, the energy used is 2.3 kWh. At 50 miles into a recorded trip with same efficiency, it shows 11.5 kWh used. The math works. So if the kWh used is also inaccurate, then the efficiency numbers are inaccurate too. I wouldn't be terribly surprised at this point, but nothing I've seen makes me think this is the case.

 

[hcsharp]

FWIW I've noticed the same behavior with our gauge. Not sure why they do it but I concluded a while ago that the gauge and range calculations on this car are for the birds. Or butterflies.

 

[philly_ev]

Honestly, the range calculator is crap. My dad had the RAV4 EV, and if his car shows 100 miles, he gets 100 miles. There is no real up or down on it.

 

[WattsUp]

Honestly, the range calculator is crap. My dad had the RAV4 EV, and if his car shows 100 miles, he gets 100 miles. There is no real up or down on it.

Are you sure you might not be exaggerating a little there? Surely there must have been some variance in the RAV4's predictions.

Ford's range calculations seem to be based a simple moving average of Wh/mi against the current SOC. It gets more and more accurate as the battery drains, and the SOC gets smaller. But, at the beginning, when the SOC is high, small changes in the average Wh/mi multiply into large changes in predicted range.

I've found the FFE's range calculations to be pretty good on flat ground with consistent driving patterns. But, they really get confused if you do a lot of hill climbing and descending. After some hilly driving once, mine predicted 150 miles of range, with like 50% battery remaining.

 

[philly_ev]

I may be exaggerating a tad bit. But I will tell you it doesn't fluctuate like mine does.

 

[hcsharp]

Honestly, the range calculator is crap. My dad had the RAV4 EV, and if his car shows 100 miles, he gets 100 miles. There is no real up or down on it.

Are you sure you might not be exaggerating a little there? Surely there must have been some variance in the RAV4's predictions.

Ford's range calculations seem to be based a simple moving average of Wh/mi against the current SOC. It gets more and more accurate as the battery drains, and the SOC gets smaller. But, at the beginning, when the SOC is high, small changes in the average Wh/mi multiply into large changes in predicted range.

I've found the FFE's range calculations to be pretty good on flat ground with consistent driving patterns. But, they really get confused if you do a lot of hill climbing and descending. After some hilly driving once, mine predicted 150 miles of range, with like 50% battery remaining.

He's not exaggerating much. If you're used to a Tesla drive train you know how well an intelligent battery gauge works. Where I live there is no such thing as consistent driving on flat ground. Does that exist anywhere? Even if most of your trip is the same speed on a highway, you still have to drive a few blocks to get to the highway, which will change everything.

BTW thanks for your theory as to how the range is calculated. I haven't been able to find that anywhere. For some reason I thought it was based on the last 6 mi of driving. If you know of any other in-depth discussions about it, point us there if you have time... Thx.

 

[danholl]

Unless the car can see into the future there is no way any car can predict the range for the next trip the moment you start out. The car can only look at the temperature to see if heating or AC is likely and consider past driving experience to predict the future mileage. Then as you drive, it refines the prediction based on current conditions. There is no magic way that can yield perfect predictions even if you are a TESLA.

One can think of things that might make it look better like always underestimate the mileage at the start and then don't adjust it till the battery gets down to 20% when one could start showing the mileage going up. For a Tesla with a long range this would not be seen on most trips.

Personally I like the way the Focus does it - making the best estimate using the data available.

 

[dmen]

Unless the car can see into the future there is no way any car can predict the range for the next trip the moment you start out.

But that's just the thing, the car has information it can use to look into the future but it doesn't use that information. That's what is so frustrating! If you program a route before starting your drive, the car knows what speeds you should be driving over what distances. If you have traffic data from Sirius, it knows where you might be driving slower than he speed limit. It knows the current temp, which is a pretty good estimate of what the temp will be in a few hours. It knows the time, so it can guess if the temp is going up or down.

And if it had just a little more information that is potentially available, like altitude along the route, actual traffic travel times, actual weather forecasts... Then it could make quite accurate predictions. Instead we're stuck with this. Sure it does a decent job with what it uses, but it could be sooo much better so easily.

 

[hcsharp]

Unless the car can see into the future there is no way any car can predict the range for the next trip the moment you start out. ...

Actually if you tell it your route then it can see into the future but that's not what I was asking of it.

What I'm asking is for the car to give me the option of calculating how many miles it thinks are remaining as if I was going to drive like the EPA does. I have not found a way to get the FFE to tell me the miles remaining based on "ideal" or "rated" range. That would usually be much more useful than what we get now. Better would be the option to have both calculations. If I know the "ideal" miles remaining, I can adjust my driving based on what I know is ahead.

For example, if I'm 15 mi from my destination, and the gauge (as it works now) says 20 miles, I know I'll be close but not much of a clue whether I'll make it or not. That's because the calculation is inaccurate (poor algorithm) and/or based on previous driving which is poor information (who knows how much elevation I gained or lost or how hard the HVAC was working?). That gauge number swings so widely that I never have a clue what the real range is. OTOH if I knew how many miles were remaining if I drove like the EPA, I could much more easily asses how I need to drive those last 15 mi to be sure I get there. I know this from driving an EV for the last 2.5 years that has both calculated (based on previous driving) and rated range. The latter is far more useful.

My wife and I like the FFE overall. Just pointing out how it needs improvement.