Published as findings. Not as claims.
Sixty-five weeks of continuous telemetry in Indian field conditions. Cell imbalance alert firmware is disabled in most deployed Indian fleets to reduce nuisance alarms — the signal is turned off. enerlytik detects what the BMS cannot. Every week. Without the alert.
The safety net was calibrated for conditions that do not exist in Indian operation.
Pack3401 degrading at −0.0077 SoH/week. Pack3001 at −0.0013 SoH/week. Same operator. Same routes. Same nominal specification. The slope difference — invisible to the BMS, invisible to any single operator, invisible to the lab — confirmed across 27 batteries, 3 cities, 13 months. Attribution to operator behaviour: less than 15%. The gap is in the design.
The field reveals what the lab is designed not to show.
14 batteries flagged as critical — all showed identical symptoms to batteries that needed replacement. None of them did. ₹5.6L avoided in one quarter by separating the charger swap from the cell replacement. Without attribution, every one of those 14 would have been replaced. Knowing when not to act is worth more than knowing when to act.
Attribution is not a feature. It is the decision — and the saving.
For LFP chemistry, the BMS SoH report is systematically decoupled from true electrochemical health. This is not a firmware failure — it is a physics constraint. The flat OCV plateau makes SoH estimation from voltage alone fundamentally unreliable.
The gap between the report and the reality is where decisions fail.
Firmware v2.1.3 was firing alerts 40mV too early, invisible to any single operator but visible across 500+ batteries. One patch. 144 service calls eliminated per year. Every false dispatch before it was a cost attribution would have prevented.
The cost is not the repair. The cost is the dispatch without the diagnosis.
Every published battery specification is derived from controlled laboratory conditions: 25°C ambient, standardised duty cycles, regulated charging. Indian field operating conditions are a different environment entirely. The BMS overstates range by 4.4km per trip — a systematic +5.5% SoC firmware bias confirmed across 179 batteries.
The specification was not wrong. It was never built for India.