TickTock said:
OK. I don't think it's a simple polynomial translation anymore. I've been experimenting with polynomials up to 7th order and cannot seem to get anything better than gids even if I split it into two sections at the knee. I think someone said that Nissan does integrate the actual charge and does "corrections" periodically. I can only imagine these "corrections" are based on the pack volts somehow.
A more targeted formula may work better than a polynomial. Any of the battery geeks out there know of any papers suggesting a formula translating charge to pack volts? It does look a bit like a natural log. The idea is if I can get a good enough formula of charge vs. voltage, we could calibrate it to current battery conditions during charging and use that to indicate remaining capacity. Can try a big look-up table but an equation would be ideal.
(Much longer than I planned on... and I'm fairly certain you've seen most of this before.)
If you assume that the LEAF battery pack works like a laptop computer battery pack, perhaps some clues fall out. Speaking towards laptop batteries (and one 5kWHr 15-cell Li Ion battery pack I've worked with)....
Most of the time, the battery voltage is fairly constant, and doesn't tell you anything about the SOC. Voltage does change with temperature somewhat (I think voltage rises as temperature rises, possibly not helpful), or when the battery SOC is near its extremes of SOC during charge and discharge. In general, battery voltage is fairly constant across capacity given a constant current (charge, open circuit, or discharge).
On charge from low SOC, current is limited and cell voltage will rise rapidly. (At very low SOC (<5%), current may be further reduced until a minimum safe voltage is reached.) At some point, cell voltage rises to a plateau and doesn't change much (15-85% SOC). Voltage starts rising again at the end of the plateau. At some voltage threshold, a timer is started. When that timer runs out, charging is stopped, and the battery is considered to be at 100% SOC. (LEAF DCFC - detecting when battery voltage climbs off the plateau may not be so easy. The plateau may have a slope indistinguishable from a normal voltage rise seen near End Of Charge at normal current. And forget the timer - stop now to be safe.)
On discharge from 100% SOC, cell voltage starts high, but rapidly falls to another plateau, and stays there until voltage begins falling again (about 15% SOC).
Falling off the plateau is where I expect the LEAF to start generating Very Low Battery warnings, and soon, enter Turtle mode to protect the battery. Entry to Turtle mode may be when the pack is considered at 0% SOC, and would be the other reference point for battery capacity.
LEAF: Have the edges of the battery voltage charge and discharge plateaus been seen? Do the plateaus have a slope? Do the plateau voltages vary with temperature?
Laptops: The strongest indicator of SOC has always been integrating battery current (counting Coulombs). A large step towards an SOC indication is finding a way to measure battery current, or finding where an integration result is stored.
The LEAF energy display apparently indicates some measure of power for traction, climate, and other loads. You would want to find where that data is coming from. Perhaps the display data is already accurate enough if it can be captured. Convert power to current for SOC.
Laptops: Corrections to SOC are required for: calendar age of the battery, equivalent full charge/discharge cycles (i.e. the sum of all energy passing through the battery), time since last charge, temperature history since last charge, and time/energy since last equalization. Most factors work to reduce calculated battery capacity. This data and techniques to implement it are likely considered proprietary.
Equalization is an interesting event. It ensures that all cells are fully charged without overcharging any cells. During a low current charge near 100% SOC, cell voltages are monitored. If any cell voltage reaches a threshold voltage, pack charging stops, and the fully charged cell is discharged slightly. After a short time, pack charging resumes, and the cycle repeats. When all cells are fully charged, the pack is at 100% SOC - this is a critical reference point. As you might guess, equalization can take a long time.
A recalibration cycle usually starts with an equalization charge, followed by a discharge to when the first cell reaches 0% SOC (voltage?) threshold. During discharge, current is integrated. When the discharge is stopped, you know what the battery capacity is. A recalibration cycle is the only time that can increase the calculated battery capacity.
When does the LEAF perform equalization charges? It's possible that cells (or modules) that reach full charge are slightly discharged while the rest of the pack continues charging. A slight rise of battery pack temperature may also be detected - discharge power has to be dissipated somehow. Do recalibration cycles ever occur? A charge from Turtle mode to 100% using the trickle charger may perform a similar function.
Laptops: Total battery capacity is influenced by age, number of C/D cycles, temperature, and time/cycles since last equalization. These constantly decrease the calculated battery capacity.
(Enough! I have other things to do.... G'nite.
)