Gary, let's take this apart. I'd like to understand what you're trying to say here.
garygid said:
Although the BMS can measure the net kWh leaving the battery since it was last charged, and could do this quite accurately,...
The car has three current sensors installed. One is in the pack and measures energy coming in and out. I haven't positively identified the other two, but have seen CAN bus messages that report regen energy flow from the inverter, and charger current, so expect the other two current sensors are in or near those systems. This suggests to me that we can get a pretty accurate count of energy movement and sources. (Let's ignore small tolerances this time around...)
garygid said:
...it has only an inaccurate ESTIMATE of the Pack's SOC, obtained from measuring the voltages on the cells, and probably using the lowest cell's voltage to ESTIMATE that cell's SOC.
I expect the phrase 'inaccurate estimate' wasn't intended to communicate 'measurement confidence'.
Here's a place I don't agree. Here's why.
Voltage isn't a direct measurement of state of charge (SOC) but we have had plenty of time with lead-acid batteries and have been able to accurately estimate state of charge from the voltage (with a temperature allowance). We've been able to do that because lead-acid has a sloped discharge curve. The red line shows a constant current discharge curve from a sealed lead acid battery. The black line is the discharge curve from a LiFePO4 battery under the same conditions.
Please note the downward slope of the lead-acid battery VS. the flat discharge of the LiFePO4 battery. We cannot measure SOC from LiFePO4 until the cell is nearly completely charged or nearly completely discharged - but we can track lead-acid's SOC throughout the entire charge or discharge.
Here's where you say "why should I pay attention to this nonsense - we don't use lead-acid or LiFePO4 in the Leaf!" - and you're right. Here's why it's important to see two examples of the SOC measurement problem:
This is a series of constant current discharge charts from a laminated LiMn2O4 cell - just like we use in the Leaf. The discharge slope is more dramatic than lead-acid.
I hope you can see that we can get a pretty good estimate of SOC from LiMn2O4 - and likely a more accurate determination than lead-acid due to the steeper slope.
garygid said:
Sure, it might calculate a 87.1% SOC from a 3.52v reading, but the actual SOC is probably +/- 5% of that "calculated" value. While driving and current out of the pack is fluctuating substantually, it is even difficult to "calculate" the cell's "resting" voltage, since each cell's internal "resistance" will vary with age, temperature, and current flow rate. Sure, averages can be used to create estimates of the "resistance", and those estimates used to calculate an estimate of the cell's resting voltage from an in-use measured voltage, and that value used to "approximate" the cell's SOC, assuming it is a cell with "typical" characteristics.
So, the calculated SOC is not accurate at all, just an approximation.
I see a couple of concerns here. First, we know that the car is aware of current into and out of the pack. We know it's aware of ambient temperature from the front bumper sensor, and of pack temperature from the four temp sensors in the pack. We know that the car's battery controller monitors and logs individual cell degradation and cell internal resistance. We don't pull energy from an EV pack the way a bench test can maintain a constant discharge current - but we also don't spent the entire 2-hour drive with our foot on the floor. There are plenty of places in the drive where we have true 'resting voltage' to measure. And since the car is aware of and actively using all of the other parameters necessary, it can continue to model and track the discharge. So no - I absolutely do not agree with what appears to be an attempt to suggest any attempt to estimate SOC while in motion is 'voodoo.'
garygid said:
Yes, you can know that you have used 13.15 kWh from the cells, but you really have no way of telling exactly how full the pack was when you started measuring, or where the "bottom" of the pack (weakest cell) will be until you actually get there.
Perhaps the "yearly" test is a "fill-drain-and-refill" test?
We've talked about this before, Gary (including the charts - I
reported this back in May) and while I agree that the "actual" SOC and the "car's SOC log" can diverge over time, I've studied the GM EV1/S10 system and the Ranger EV system. I think the Leaf's battery monitoring system is a significant improvement over the other two - and over ANY of the monitoring or management systems available to the do-it-yourself market. While I desire truth more than ego or post count, and will absolutely accept being proven wrong when we get more real-world experience with the car, I don't think our SOC numbers are going to be that far off. Pizza bet time?
I've not seen anything in the manuals that suggest that the annual pack test includes either a discharge or recharge. The car's systems track each cell and the pack health report can be generated in seconds by downloading the data from the car. The only time cells are charged or discharged is when cells and/or modules are replaced in the pack.
SOC is directly reported and available on the CAN bus, as are cell condition variables like internal resistance, voltages, pack temperature and the rest. Time will tell.
edit...fixed link