Insights / Field note · BESSBy Robert Sills CEng MIMechE2 June 2026Updated 9 August 2026

Why state of charge is so hard to estimate in a BESS

State of charge is one of the most important numbers for a battery asset and one of the hardest to measure accurately.

One of the main challenges in measuring state of charge (SOC) accurately is how flat the open circuit voltage (OCV) curve is across the normal operating range.

Line chart of open circuit voltage against state of charge for an LFP battery cell, showing a flat curve between 10 and 90% with only 141 mV of change and steep ends inside the control reserve bands. — Measured OCV against SOC for a representative LFP cell. Across the 10 to 90% operating window the voltage moves just 141 mV, about 4.4%, which is why voltage alone is a weak SOC signal.

What the data shows

This plot, from measured data for a representative lithium iron phosphate (LFP) cell, shows that across the normal operating window of 10% to 90% SOC, the cell voltage changes by only 4.4%.

Why voltage alone is not enough

Combine that small signal with voltage measurement error, temperature shifts and different curves for charge and discharge, and direct voltage based SOC becomes unreliable.

What is used instead

Instead, SOC estimates are based on integrating current from a known point, known as Coulomb counting. The estimate is then re-calibrated at high and low SOC, where the OCV curve is steep and the reading is more accurate.

Method notes. The curve is average open circuit voltage against state of charge for a representative lithium iron phosphate cell at 25 °C, taken from the CALCE battery dataset. It is the mean of the charge and discharge curves, so the hysteresis between them is not visible here and widens the real uncertainty further. The 141 mV span is read between 3.209 V at 10% SOC and 3.350 V at 90%, and the 4.4% is that span relative to the lower voltage. Chemistry, cell age, temperature and current all move the curve, and the operating window and control reserve bands shown are indicative and vary by owner, OEM and warranty. A pack behaves differently again, because cell-to-cell spread and the balancing strategy sit on top of this single cell picture.

For owners, undetected SOC drift means trading decisions are built on a number that is quietly wrong.

Charts and analysis by PowerVeritas. Where open datasets are used, sources are credited on the attributions page.