What is a "Gid"?

TickTock,
Have you continued doing some 100% charging sessions with at least several hours afterward plugged in?

If so, have your max-GIDs kept climbing?

If so, it might be that you got a car that was WAY out of balance, perhaps due to one module being replaced during post-manufacturing quality control.

evnow said:

The unpredictability of gids w.r.t. usage is one reason I've almost stopped using my SOC meter - I no longer log it after every trip.

Click to expand...

Perhaps for you, but I find it to be very predictable. I live in Southern California where temps don't vary a lot. My round trip commute of 45 miles went from 224 gids (80%) to 100-105 gids during the summer. In winter, it is a bit lower as I tend to arrive home with 90-95 gids. Of course, if you have a variation in elevation, stop and go driving, more aggressive acceleration, large temperature changes, etc. your results will be different.

Ingineer said:

Nissan calls it stored watt-hours... Either way, I can assure you what you are calling SOC, Nissan is calling stored watt-hours.

Click to expand...

TickTock said:

Since gids are a unit of energy and the battery has losses, gids out will always be less than gids in. Total round trip efficiency is 85% and half of this comes close to 75/80 so the numbers start to make sense if you account for losses. The 281 everyone (else) sees after a 100% charge is probably "gids in." Not to be confused with "gids out." Add this on top of the integrating error, hall effect sensor drift, etc, and we can see why the ECU needs to adjust the value time to time.

Click to expand...

I agree with what Luke and Phil said above. There is a difference between stored and expended energy. I'm not aware of any empirical data that would support more than 75 Wh per gid when looking at the expended energy. Luke's plot earlier in the thread was recorded when charging, not when driving. His findings support 80 Wh per stored gid.

I have another datapoint for that. The general agreement seems to be that the difference between a 80% and a full charge is 4 Kwh. Measured from the wall and adjusted for charging inefficiency. I used Gary's meter for much of last month, and a 80% charge corresponded to 231 gids. A full charge was either 280 or 281 gids. A delta of 50 gids, 4 kWh, or 80 Wh per gid stored.

I tried to validate this number in the spreadsheet I compiled for the reverse SOC meter project, and the best I got was 3.67 kWh or 73.4 Wh per gid expended. I chalked it up to low temperature and losses, although I still don't understand where exactly they come from. Drivetrain inefficiency should be baked into the MPK number, and the coulombic efficiency of the battery is likely greater than 99%. What is the accuracy of the odometer? What else could we be missing?

80 Wh into the battery while charging (not from the wall, which would be substantially more Wh) MIGHT be counted as an increment in the GID counter.

80 Wh in is perhaps 77 actually stored, and then 73 Wh actual usable energy when recovered from the battery.

That, plus or minus a bit, would fit a lot of the data.

But, while L2 charging, assuming (might be wrong) a more or less "constant" current into the battery (difficult to tell because of the huge noise), the energy rate (watts) would be increasing as the pack voltage increases. That would make GIDs rise faster in the later parts of the charge. Does anybody see that increase (I have not)?

If the charging current drops off to achieve a constant-watt charging rate, the current would drop and the GIDs would increase linearly. That is possible, but I have not seen the current taper off, and a roughly 25% decrease should be noticable.

Maybe I really need to plot charging current vs battery pack voltage, or use a real "time" axis, etc.

I need to improve the data-plotting functions in CAN-Do a bit more.

garygid said:

If the Gids are essentially 80 watt-hours each, why are the top 20 or so "always" used up so much quicker than those in the center of the range?

Click to expand...

Less regeneration when the battery is full? Not enough details regarding your observation.

At least now I know what's a "gid"

Any idea on the type of Hall effect sensor used in the LEAF? I would have preferred they use a MR device similar to http://fwbell.com/Page.aspx?Page=11" onclick="window.open(this.href);return false;

PDF is http://fwbell.com/ckfinder/userfiles/files/NT-Datasheet.pdf" onclick="window.open(this.href);return false;

Stoaty said:

lukati said:

For me the important information coming from Phil's data is the fact that the BMS protects only 2% capacity at the bottom and ~5% at the top. Unlike the Prius NiMH battery that Toyota won't let you abuse, your use pattern on the LEAF will truly affect the longevity of the pack. We knew that going in, now we have the numbers to prove it.

Click to expand...

The 5% at the top doesn't surprise me, but I am stunned by the 2% at the bottom. These numbers just reinforce that I want to continue using the middle 60% most of the time for maximum battery longevity. Of course, we won't know exactly how much difference the usage pattern makes for a few years until Leafs start dropping like, well, Leafs.

Click to expand...

I am with you on this. I am trying (for the most part successfully even in cold Chicago) to fit my daily 50 miles into the middle 60%. But this will make me more cautious at the bottom end. It needs to be said, however, that battery capacity and SOC are an inexact science. Different manufacturers use different criteria for calculating capacity. So Nissan's 2% SOC may be more conservative that someone else's 4%.

surfingslovak said:

I agree with what Luke and Phil said above. There is a difference between stored and expended energy.

Click to expand...

It is true that not all the stored energy makes it to the wheels.. but that really is not important for us, all we want is a consistent range when the car is driven on a consistent basis within a small tolerance of 5% or so. All we need is a GOM with an instant type reading, and another one with a long term average of a couple of weeks or so... perhaps that would be too complicated for most people.

My guess is Nissan does not take the losses into account when charging so when you start driving it corrects the gid value based on the battery voltage and ??? other inputs. This could explain why the gids drop fast at first. Anyone able to confirm observing this fast drop in the gid count after a *80%* charge? If my theory is correct, the phenomenon should exist after both 80% and 100% charges and not be tied to actual battery SOC.

I agree that the gids are probably a capacity meter and not a SOC. I noticed that sometimes the gidometer only gives 260gids after a full charge, and in this cases the SOC bars disappears at a lower threshold than when the charge is at 100% (The first bar dissapears usually at 255 gids, but when a 100% charge stops at 260 gids it disappears at around 235 gids). This is consistent to the BMS thinking there is a lower capacity in the battery and lowering the threshold for each SOC bar drop. When this lower assumed capacity happens it is usually corrected after the car is power cicled and the gids go up.

In my measurements the gids are a low confidence value of the true battery capacity. I find that the best indicator of SOC is the OCV. The kW.h needed to charge the battery to 100% are much better correlated to the initial OCV than to the initial gid. For the same initial charging gid there is a spread of more than 1,5kW.h in the actual used energy to charge to 100%, but less than 0,5kW.h spread for the same initial charging voltage. I don't know if the cause is a bad estimation of the SOC or a cell balancing problem and the BMS is giving the remaining usable capacity adjusted for the weakest cell.

vegastar said:

In my measurements the gids are a low confidence value of the true battery capacity. I find that the best indicator of SOC is the OCV.

Click to expand...

What is OCV?

Open Circuit Voltage wich is also given by Gary's SOC-meter (for the true OCV the car must be in ACC mode, with the main contactor open).

vegastar said:

Open Circuit Voltage wich is also given by Gary's SOC-meter (for the true OCV the car must be in ACC mode, with the main contactor open).

Click to expand...

Thanks. Are you suggesting that checking the OCV after a 100% charge would be a good proxy for battery capacity and that it will go down over time as the battery ages? If so, I am going to record mine and follow it over time.

garygid said:

80 Wh into the battery while charging (not from the wall, which would be substantially more Wh) MIGHT be counted as an increment in the GID counter.

80 Wh in is perhaps 77 actually stored, and then 73 Wh actual usable energy when recovered from the battery.

That, plus or minus a bit, would fit a lot of the data.

But, while L2 charging, assuming (might be wrong) a more or less "constant" current into the battery (difficult to tell because of the huge noise), the energy rate (watts) would be increasing as the pack voltage increases. That would make GIDs rise faster in the later parts of the charge. Does anybody see that increase (I have not)?

If the charging current drops off to achieve a constant-watt charging rate, the current would drop and the GIDs would increase linearly. That is possible, but I have not seen the current taper off, and a roughly 25% decrease should be noticable.

Maybe I really need to plot charging current vs battery pack voltage, or use a real "time" axis, etc.

I need to improve the data-plotting functions in CAN-Do a bit more.

Click to expand...

This confirms my assertion that "gids" are watt-hours, not SOC or Amp-hours. Again, the wH/mile figure is probably pretty rough, but the SOC and stored Watt-hours are probably more accurate. All this is confirmed from my own findings in both the Leaf, and in my years of experience designing battery management systems, including one very similar to the Leaf's that is installed in my Plug-in-Prius.

Since my pack in the Prius is small, I don't bother correcting for the slight hall-effect drift except at the ends, where it matters. I think the Leaf's BMS does this as well, which is why we see sudden inaccuracies at the top and when running it into the ground.

-Phil

Stoaty said:

vegastar said:

Open Circuit Voltage wich is also given by Gary's SOC-meter (for the true OCV the car must be in ACC mode, with the main contactor open).

Click to expand...

Thanks. Are you suggesting that checking the OCV after a 100% charge would be a good proxy for battery capacity and that it will go down over time as the battery ages? If so, I am going to record mine and follow it over time.

Click to expand...

No. The OCV is a good proxy for the current SOC, but not suitable for the total capacity. As the battery ages the 100% SOC OCV will be the same 4,10V/cell or 394V total pack voltage. What I am doing is checking for the same OCV if the energy needed to charge to 100% SOC is diminishing. As an example, I usually end the day with around 30% SOC and 363V OCV. To charge to 100% from 363V I need around 17,5kW.h. If the capacity diminishes by 10%, then for the same 363V of OCV I will need only 15,75 kW.h measured from the wall.

The best way to check battery capacity is to discharge it completely and check the energy from to wall to charge it. Knowing the OCV before each charge is a easier way to keep track of the capacity without the need to discharge the battery completely.

It would be nice to correlate OCV values with GID values, with the condition that the pack starts with a 270+ GID reading on a 100% charge...

Could be time consuming - how long does it generally take for the battery to rest before OCV stabilizes?

drees said:

It would be nice to correlate OCV values with GID values, with the condition that the pack starts with a 270+ GID reading on a 100% charge...

Could be time consuming - how long does it generally take for the battery to rest before OCV stabilizes?

Click to expand...

This would also require knowing the exact cell temperature, as it changes with temperature.

-Phil

drees said:

It would be nice to correlate OCV values with GID values, with the condition that the pack starts with a 270+ GID reading on a 100% charge...

Click to expand...

Yes, this is something I started doing last month while working on the reverse SOC meter project. I wanted to correlate battery bars to pack voltage and gids. I collected quite a bit of data, but post-processing can be time-consuming.

Perhaps it would be worth logging it from the CAN bus, using Gary's meter for example. Any errors would average out across samples, which would possibly help sidestep the issue of temperature and other environmental factors. This worked relatively well for me in the last few months, and the data I've collected, even though imperfect and error-prone, generated acceptable results.

Herm said:

It is true that not all the stored energy makes it to the wheels..

Click to expand...

Herm, thanks for your comment. I think that devising either a more accurate or at least a more practical way to predict range is a shared goal for many of us.

I'm curious what your thoughts on the coulombic efficiency of the battery pack were. Are heat losses included in that number? I believe that heat generated in the pack is one of the items we have overlooked so far. Unfortunately, the corresponding energy loss is likely less than 0.5 kWh for a full charge, and it won't explain the 8% delta between energy in and energy out.

surfingslovak said:

I'm curious what your thoughts on the coulombic efficiency of the battery pack were. Are heat losses included in that number?

Click to expand...

All the losses in the battery turn up as heat... supposedly the batteries are 99% in a zone around the 50% SOC, and I have seen reports of 95% efficiency, perhaps at a wider SOC. You also have motor/inverter losses and gearing losses.

Herm said:

You also have motor/inverter losses and gearing losses.

Click to expand...

Yes, but wouldn't these be baked into the MPK (energy economy) number? I can understand that the MPK figure is rough, as Phil said earlier, and perhaps not even very accurate, but a consistently lower reading and a delta of 8% cannot be attributed solely to instrument error. I believe that the Leaf is measuring apms out of the pack, and it tracks pack voltage. Every Watthour that leaves the pack should be accounted for. Ergo, the most likely culprit is heat development in the battery and coulombic inefficiency. Or did I misunderstand something?