Page 4 of 7

### Re: Real SOC (State of Charge) - Information

Posted: Mon Mar 19, 2012 6:34 pm
Ingineer wrote:All we need to calc kWh remaining is the total pack amp-hour figure (67.568 on my car) and then correct that for the actual usable range, then multiply that by the SOC percentage to get amp-hours remaining, then multiply that by the current average voltage. Or, just display amp-hours directly and use that.

-Phil
That would assume that the average voltage is stable enough over the duration of the trip. Counting amp-hours should be easy enough, but amps alone dont move the car...Don't we need the actual voltage too, which can vary considerably ? Or maybe it doesnt?

### Re: Real SOC (State of Charge) - Information

Posted: Mon Mar 19, 2012 6:51 pm
klapauzius wrote:That would assume that the average voltage is stable enough over the duration of the trip. Counting amp-hours should be easy enough, but amps alone dont move the car...Don't we need the actual voltage too, which can vary considerably ? Or maybe it doesnt?
The voltage isn't stable over the trip, it fluctuates and the short term average slowly falls. You don't count the amps, you let the car do it (this is how it knows SOC) and then the voltage is averaged over a short period (to prevent momentary spikes from altering your watt hours) and then multiplied by the calculated amp-hour figure.

I'll add it to LEAFSCAN and test it.

-Phil

### Re: Real SOC (State of Charge) - Information

Posted: Mon Mar 19, 2012 7:03 pm
Ingineer wrote: What I'm saying is I don't want to try to make another inaccurate range display, I'll leave that calculation to the wetware upstairs. I want to give that wetware as much accurate information as I can at high resolution and at a fast rate.
We are all different, but I would like to see a simple remaining range estimator based on my average m/kWh for the last week, programmable by each user. If SOC provided by the Leaf itself is the most accurate then base the calculation on that...

If you drive so many miles per day, and you use a certain % of the SOC doing that then each driver can work out their own personal range parameter.

### Re: Real SOC (State of Charge) - Information

Posted: Mon Mar 19, 2012 7:12 pm
klapauzius wrote:I wonder if that ever has been corroborated by a real scientific study? While the range estimation is not very accurate at high levels of charge, once you have driven the car a bit, it becomes in my subjective (n-of-one) experience, quite accurate.

Again with automatic logging (or even pen and paper) it should be trivial to sum up actual miles driven + DTE and check how constant these numbers are over the length of the trip. If we do that across e.g. 100 drivers in different locations, we can easily compute a confidence level for the GOM and conclude if it is really dog-doo-doo or the innocent victim of prejudice...
I'm not sure if this qualifies as a scientific study, but as I've mentioned here and elsewhere, I have done fairly extensive range testing for the reverse SOC meter project. To be honest with you, I was also getting tired of the constant second-guessing and kibitzing by folks that have rarely, if ever, driven their Leaf to turtle (and beyond). Tony has led the charge on this front, and I trust his judgement. He has been nothing but methodical. I have several hundred data points as well, and although I don't have much time for the requisite postprocessing, I can assure you that the GOM is borderline unusable. It consistently overestimates the total range by 15 to 30% on a full charge, and it underestimates the true range by up to 50% as you approach the low battery warning. It's fairly accurate in the middle of the range, which demonstrates that it can work too, but it fails to do so on either end of the SOC range. This was measured over several full battery cycles, with consistent driving and weather patterns on flat roads. In other words, ideal conditions.

I don't expect miracles, just something that works reasonably well, or I can configure to make work reasonably well for my needs. If Nissan allowed me to insert an energy economy factor that would be multiplied by the estimated available energy, I could stop doing the mental math. I would not mind if the vehicle used an aggregate m/kWh factor, the one that's displayed on the dash for example, which would get reset every time I reset the gauge. That way, I would have control over the integration interval. It could be minutes, hours, days, weeks, moths or even the life of the vehicle.

Speaking of which, I think that it's insane for Nissan, BMW, or any other manufacturer, to adjust the full range prediction based on a few seconds of driving. Absolutely insane. You get erratic behavior like we all know too well from the Leaf, or like I saw in the ActiveE this morning. My total range estimate jumped from 150 miles to 190 miles after a few seconds of regen. Seriously? The range prediction should be fairly consistent, constant, and based on a longer time interval. Hours, days even. If you wanted to alert the driver that current road conditions or driving patterns will severely impact range, you could show a message or an alert, much like we see with A/C in the Leaf: if you keep driving like this, your total range will decrease (or increase) by XX miles.

Basically, you would work with two range estimates, one based on a medium to long-term integration period, and another one based on a short-term integration period. You would display the more constant of the two figures, and the delta. If you cared to achieve the projected range, you would want to keep the delta close to zero. Likewise, if you were running low, and needed to conserve energy, you would want to keep the delta in positive territory.

Wouldn't that be good enough?

### Re: Real SOC (State of Charge) - Information

Posted: Mon Mar 19, 2012 8:43 pm
surfingslovak wrote:Basically, you would work with two range estimates, one based on a medium to long-term integration period, and another one based on a short-term integration period. You would display the more constant of the two figures, and the delta. If you cared to achieve the projected range, you would want to keep the delta close to zero. Likewise, if you were running low, and needed to conserve energy, you would want to keep the delta in positive territory.
That would be fabulous. Particularly for people like myself who use the car primarily for commuting and thus drive the majority of miles on the same route. In a place like Chicago, you would probably want a rolling average over a two-three week period to take into account the seasonal temperature variation. On particularly cold days you would know that you have to keep the delta positive in order to achieve the long-term average range.

### Re: Real SOC (State of Charge) - Information

Posted: Mon Mar 19, 2012 10:24 pm
klapauzius wrote:
I wonder if that ever has been corroborated by a real scientific study? While the range estimation is not very accurate at high levels of charge, once you have driven the car a bit, it becomes in my subjective (n-of-one) experience, quite accurate.

Again with automatic logging (or even pen and paper) it should be trivial to sum up actual miles driven + DTE and check how constant these numbers are over the length of the trip. If we do that across e.g. 100 drivers in different locations, we can easily compute a confidence level for the GOM and conclude if it is really dog-doo-doo or the innocent victim of prejudice...
I reset my A trip meter after every charge and watch the sum of DTE and TripA which with an 80% charge starts out at 88 and after 5 mins of 25mph streets jumps to 92 and then slowly degrades for the rest of the trip including 9 miles of 65+ freeway driving to a net of 65miles. I'm hoping this type of analysis will be semi automatic with the tool Ingineer is making.

### Re: Real SOC (State of Charge) - Information

Posted: Tue Mar 20, 2012 1:10 am
TickTock wrote:Cetainly a good way to get confidence. However I don't think it is necessary. For my car, which is an outlyer, I get 80wh/gid. My max gids of 245 versus 281 for everyone else correlates really well to the 22kWh I measure from the wall versus 25kWh that everyone else sees. I had already estimated a gid=80wh for my car by logging the battery votlage and amps for several trips before Phil indicated that this was the standard. It is also very linear across the entire battery range. So, while you could go through all these steps to get better accuracy, simply multiplying by 80 should be good enough.
The battery ECU sends the watt-hours figure out over the CAN bus so the range calculations can be done, and I can say with 100% certainty that figure is Watt Hours/80. This is what everyone is calling the "GID". All Leafs use the same figure, but not all Leafs contain the same battery capacity, and not all Leafs charge to the same amount each time, which is why the watt-hour figure varies. This would be a great parameter to monitor for determining available charge, except it suffers from a low resolution.

Personally I identify better with a 0-100% number than 0-22480 (or less). I understand that as the car ages this will mean 100% has less range than it used to, but as long as you have access to the full capacity figure, or the loss coefficient (and we will have both), it still makes sense to me.

-Phil

### Re: Real SOC (State of Charge) - Information

Posted: Tue Mar 20, 2012 7:06 am
"Ingineer"
...Personally I identify better with a 0-100% number than 0-22480 (or less). I understand that as the car ages this will mean 100% has less range than it used to, but as long as you have access to the full capacity figure, or the loss coefficient (and we will have both), it still makes sense to me.

-Phil
Phil-

Agree, except, as I asked yesterday:
What variability of battery capacity are you seeing, with battery temp changes only?

I don't think capacity loss over time will be too great a problem for a driver watching a % of capacity display, as an indication of available range, since this will occur slowly, over years.

But if battery capacity changes significantly due to battery temp, the % (estimates are all over the place for this AFAIK) could be misleading for range purposes, from day to day, or even over a single day, even for drivers capable of using "wetware".
And, I think the ability to see the display calculated as "kWh remaining" as well as a "% of available capacity" will be the "range" information, most drivers will want.

### Re: Real SOC (State of Charge) - Information

Posted: Tue Mar 20, 2012 8:03 am
Ingineer wrote:
TickTock wrote:...

Personally I identify better with a 0-100% number than 0-22480 (or less). I understand that as the car ages this will mean 100% has less range than it used to, but as long as you have access to the full capacity figure, or the loss coefficient (and we will have both), it still makes sense to me.

-Phil
The nice thing about watt-hour display is that I can relate it to my expected mi/kWH or conversely, try and adjust my mi/KWH to fit my fuel budget. Of course I could do the additional math but it seems much more direct and intuitive for me to start with units of energy.

### Re: Real SOC (State of Charge) - Information

Posted: Tue Mar 20, 2012 8:08 am
Many of the operational parameters (like Pack Amps) have a lot of short-term variation (or "noise"), and require various "smoothing" (usually averaging) time constants to get "viewable" data.

When the car is ON and not moving, there should not be any IN (charging) current to the pack, but sometimes the SOC-Meter indicates under an amp of "charging".

Perhaps its 20-sample "averaging" has a bug, or the car's current sensor is not accurate enough around zero?