BESS module imbalance: what voltage spread can and cannot prove

What the evidence says

Cells connected in series share current, so the first cell or module to reach an operating limit can constrain the string. That observation explains why imbalance matters, but one screenshot near the top or bottom of charge does not reveal whether the cause is capacity dispersion, resistance, temperature, measurement offset, state-estimation drift or balancing behavior.

Voltage comparison only becomes interpretable when state and timing are controlled. Readings under load include resistance and connection effects; readings immediately after charging include relaxation; temperature changes open-circuit voltage and usable power. A diagnostic plan therefore compares repeatable charge, rest and discharge segments instead of mixing arbitrary moments.

Thermal gradients and current-sharing differences can create non-uniform ageing. Cooling paths, rack position, loose or resistive connections, contactors, fuses and parallel-string impedance must be examined beside the BMS record. Replacing the lowest-voltage module without checking these system conditions can reproduce the symptom.

Balancing is bounded by the BMS design, permitted state, balancing current and available dwell time. It cannot repair lost active material, every high-resistance fault or a bad sensor. A long balancing session is not proof of recovery unless a repeatable capacity and power test shows that the limiting spread and delivered result improved.

What a business should do next

1. Preserve original logs before changing limits, firmware or modules. Export the highest available resolution for cell or module voltage, temperature, rack current, SOC/SOH, balancing state, alarms, contactors, charge and discharge limits, PCS power and EMS commands.
2. Verify measurement integrity with calibrated independent instruments and connection inspection under a controlled safety procedure. Record timestamps, sensor identifiers, firmware, ambient conditions and any data gaps so apparent imbalance is not confused with instrumentation error.
3. Rank hypotheses and attempt the least invasive reversible test first. Temperature normalization, connector checks, controlled rest comparison or documented balancing may precede replacement, but any intervention must respect manufacturer limits and site safety responsibilities.
4. Close the work with a witnessed reference performance test under stated starting SOC, ambient, power, meter boundary and pass/fail criteria. Compare usable AC energy, limiting cell or module behavior, temperature spread and alarms against the pre-intervention baseline.