This simulator exposes the hidden true state of charge of a generic 3 Ah NMC-style teaching cell, alongside a coulomb-counting estimator that a real battery management system would have to rely on instead. Introduce a current-sensor bias or a wrong starting estimate and watch the estimate drift away from the truth — then use a rested-voltage correction to bring it back.
• A real-time 3D fixture with the cell module, a four-terminal current shunt, a current integrator/coulomb counter, a true-charge reference gauge, an estimated-charge indicator gauge, and a discharge/charge fixture — with home view, focus-selected-part, full-enclosure toggle, exploded view, auto-rotate, expand and label controls. • Experiment controls for initial SOC (new trial, the hidden truth), pack current from −6 A charge to +12 A discharge, ambient temperature, initial estimated SOC (new trial) and current-sensor offset (−0.2 to +0.2 A). • Live readouts for model SOC, estimated SOC, estimation error (percentage points), actual current, loaded voltage and net charge out (Ah), plus an operating-sequence readout, switch-state panel and cell-readings table. • Playback controls: pause/resume, advance 1 s or 60 s, and a speed selector from real time up to 1 hour per second. • An Analysis tab with two charts and the estimator equations (SOCestimate(t) = SOCestimate(0) − ∫(I+bias)dt/(3600Q), Error = SOCestimate − SOCmodel, and the rested OCV⁻¹ correction) plus snapshot measurements. • A Test & diagnose tab with four guided experiments (a biased gauge, an initial-estimate error, rest-and-calibrate, and reversing to charge direction), a model-verification bench, and a timestamped event log with report export. • A Learn & assess tab covering SOC as a hidden state estimate, coulomb counting and its sensitivity to sensor bias, open-circuit-voltage correction, and how practical fuel gauges combine multiple estimation mechanisms, plus a two-question knowledge-check quiz and a written model-scope statement.
A coulomb counter estimates SOC by integrating measured current over time, starting from an assumed initial SOC and a known cell capacity. Any constant current-sensor offset gets integrated right along with the real current, so the estimation error grows steadily over time even when the sensor is only slightly miscalibrated — the simulator's "biased gauge" experiment applies a +0.1 A offset so you can watch the estimate fall away from truth. And because the integrator only tracks changes, it can never correct a wrong starting value on its own: an inaccurate initial estimate persists unless something anchors it back to the truth.
The simulator's "Calibrate from rested voltage" action inverts the cell's open-circuit-voltage curve to recover SOC from a measured terminal voltage — but only once the cell has actually rested. The calibration action is blocked while current is flowing or while polarization voltage remains above about 5 mV, since loaded or recently loaded voltage reflects resistance and polarization on top of the equilibrium OCV, not the equilibrium OCV itself. This threshold (|I| ≤ 0.05 A and |Vp| ≤ 5 mV) is a modeling convenience, not a universal practical rest-time criterion — real battery management systems combine electrical models, temperature and learned capacity rather than relying on either mechanism alone.
No. Coulomb counting integrates changes in charge, so it needs a correct starting SOC to begin with. If the initial estimate is wrong, that offset persists through the discharge unless a separate correction — like a rested-voltage calibration — anchors the estimate back to a known value.
The rested-voltage correction inverts the open-circuit-voltage curve, which only applies at electrochemical equilibrium. Loaded or recently loaded voltage includes an ohmic drop and polarization voltage on top of the true OCV, so calibrating from it would produce an inaccurate SOC. The simulator blocks calibration until current and polarization voltage both fall below small thresholds.
Since the estimator integrates measured current (including any constant offset) over time, even a small sensor bias accumulates into growing estimation error the longer the cell discharges or charges. The simulator lets you set the offset directly and watch the error readout grow.
This is a generic 3 Ah NMC-style teaching cell with coulombic efficiency fixed at one and no aging, diffusion, hysteresis or lithium-plating effects. The rested-voltage calibration threshold is a modeling convenience rather than a certified rest-time specification, and real fuel gauges combine several estimation techniques together with temperature and learned capacity.