Performance model validation

Constant current validation

Charge, five temperatures by four rates, one parameter set

model prediction lab measurement

Constant current charges from 0.5C to 5C at 0 to 50 °C, plotted as cell voltage against time, including fast charge rates in the cold. Every panel comes from the same parameter set. Hover any curve for values, click any panel to enlarge it below with a temperature view.

Discharge, five temperatures by five rates

model prediction lab measurement

Full discharges from 0.5C to 10C across the same temperatures, down to the cold high rate corner where the voltage response is hardest to get right.

Drive cycle validation

Constant current tests are clean but batteries rarely live that way. The WLTP automotive drive cycle exercises the model against fast changing loads.

WLTP drive cycle at three starting states of charge

model prediction lab measurement

Current mode runs at 25 °C. At mid and high state of charge the voltage tracks within about 13 mV RMSE. The 7 percent run is the stress case, where the cell rides the steep end of its voltage curve. Click any panel to enlarge it above.

Accuracy in numbers

Each dot is one constant current test. Voltage error stays in the tens of millivolts across the whole temperature and rate window.

Voltage accuracy, every condition on one axis

one constant current test

Median voltage error is 29 mV RMSE across the 45 constant current tests, with the worst at 81 mV on a cold discharge. Temperature is predicted within 1.1 °C RMSE on average. Hover any dot for its condition.

Full metrics per condition. Click a column header to sort.

Drive cycle runs, in current controlled and power controlled mode:

Version fe1f5cb5