Topic 18 - Unlocking the Potential of Advanced Charging Methods

Advanced Charging Algorithms and BMS Integration · Battery Shortcut

Section 5 · Topic 18

Advanced Charging Algorithms and BMS Integration

The CC-CV algorithm is the foundation, but driving high current into a cell without plating it requires charging that responds to the cell’s internal state, not to static voltage triggers.

Constant power and beyond

Constant-power charging limits the product P = I × V instead of current alone: at low state of charge, where voltage is low, the charger pushes high current, and as voltage rises the current tapers to hold power constant, compressing the early charge time. The gain is real but bounded by the same plating limit — which is why the more capable approach is model-based.

Model-based (adaptive) charging

Adaptive charging estimates the cell’s electrochemical state in real time and adjusts current against it.

CHARGER CELL SENSORS V · I (Kelvin shunt) T (NTC β≈3950 K) MODEL ECM + EKF / UKF estimate internal state charge current I measure V, I, T dynamic I limit Li at surface accumulating faster than it intercalates? → cut I, prevent plating
Fig. 5.18 — The adaptive loop. An equivalent-circuit or physics model, fed by live V, I, and T, predicts when lithium is accumulating at the anode surface faster than it can intercalate — the precursor to plating — and commands the charger to cut current the instant that condition approaches, holding the cell at the edge of its plating limit rather than backing off conservatively.

Pulse charging proposes to push this further by interleaving high-current pulses with brief rests, on the theory that the rests let electrolyte concentration gradients relax; the mechanism is plausible but its net benefit over well-designed continuous tapering remains contested, and it is presented as a debated modality rather than established practice. Temperature-compensated charging wraps these in hard limits: below ~10 °C the BMS caps current until the cells warm, and above ~45 °C it blocks charging until cooling restores the window.

The sensing hardware

These algorithms are only as good as their measurements. Pack current is read through a low-tolerance shunt with a four-wire Kelvin connection — sensing voltage on separate leads from those carrying current, which removes lead resistance and yields micro-ohm accuracy — often backed by a Hall-effect sensor for galvanic isolation. Temperature is sensed by 10 kΩ NTC thermistors (β ≈ 3950 K) at cell terminals and busbar hotspots. On the power path, the BMS switches the main contactors through a pre-charge resistor that limits inrush current into the pack capacitance — without it, the surge would weld the contacts — and monitors isolation resistance to catch a chassis fault.

EKF vs UKF

The two estimator variants handle the nonlinearity of the voltage–SoC curve differently: the EKF linearises the model around the current operating point, while the UKF propagates sample points through the full nonlinear model — more computation for better accuracy where the curve is sharply nonlinear, as at the ends of the LFP plateau.

Section 5 · Advanced ElectrochemistryTopic 18 / 23

Comments

Post a Comment