Second-Life EV Battery Field Data Programme

Published degradation data for EV packs comes almost entirely from vehicles. There is very little public field data on what those same packs do once they are installed as stationary storage. This programme is an attempt to build that dataset.

Status as of 2026-09-19: methodology published, dataset not yet published. This page defines what will be collected, how it will be anonymised and what will be released. No aggregate results are published yet, because the consented sample is not large enough to release without risking re-identification or drawing conclusions from too few installations. This page will be updated when that changes.

Why this data does not exist yet

Vehicle-fleet degradation studies are reasonably well covered — Recurrent and Geotab both publish large-sample analyses, and academic work such as Preger et al. (2020) characterises cell-level ageing under controlled cycling. What none of them covers is the second-life stationary case, where the duty cycle is completely different:

Depth of discharge in a residential storage system is typically 15–25 % per day, against far deeper and more variable discharge in vehicle use.
C-rate is typically an order of magnitude lower — a 75 kWh pack feeding a 10 kW inverter runs at roughly 0.13 C, against automotive peaks well above 2 C.
Thermal environment is a garage or utility room, not a vehicle with active conditioning tied to driving.
Calendar age at commissioning is already 5–10 years, so the pack starts its second life on a different part of the ageing curve.

Extrapolating vehicle degradation curves onto that duty cycle is an assumption, not a measurement. The purpose of this programme is to replace the assumption with data.

Data schema

Each contributed installation record uses the following fields. Fields marked required are the minimum for a record to be usable; the rest improve the analysis where available.

Field	Type	Required	Notes
`battery_family`	string	Yes	e.g. "Tesla Model 3 LR NCA" — matches a profile in batteries.csv
`battery_chemistry`	enum	Yes	NCA / NMC / LFP
`pack_capacity_nominal_kwh`	number	Yes	As-new nameplate capacity
`pack_year_of_manufacture`	year	Yes	Used to derive calendar age at commissioning
`soh_at_commissioning_pct`	number	Yes	Measured, with the measurement method stated
`soh_measurement_method`	string	Yes	e.g. UDS diagnostic, ScanMyTesla, LeafSpy, OEM BMS report
`inverter_family`	string	Yes	Matches a profile in inverters.csv
`inverter_firmware`	string	No	Version string as reported by the inverter
`bms_ev_hardware_revision`	string	Yes	e.g. HW3.1
`bms_ev_firmware`	string	Yes	e.g. 16.5.0
`commissioning_month`	YYYY-MM	Yes	Month only — day is dropped during anonymisation
`climate_zone`	enum	Yes	Köppen classification group, not country or city
`enclosure_type`	enum	No	Indoor conditioned / indoor unconditioned / garage / outdoor enclosure
`typical_daily_dod_pct`	number	No	Averaged over the reporting period
`cumulative_throughput_kwh`	number	No	Total energy through the pack since commissioning
`equivalent_full_cycles`	number	No	Derived from throughput and usable capacity
`soh_current_pct`	number	No	Latest measurement, same method as commissioning where possible
`cell_imbalance_mv`	number	No	Max minus min cell-group voltage at mid-SoC
`fault_events`	number	No	Count of BMS-reported protection events over the period
`reporting_period_months`	number	Yes	Operating months covered by this record

Anonymisation approach

No customer identifiers — no name, address, company, email or order reference enters the dataset.
No exact location — climate zone (Köppen group) only, never country, region or postcode. Country is dropped because with small samples a country plus a battery family can identify an installation.
Date granularity reduced to month for commissioning, and to quarter for any event timestamps.
Minimum cell size before release — no figure is published for any subgroup with fewer than 5 installations. Subgroups below that threshold are merged upward or omitted.
Opt-in only — a record enters the dataset when the owner explicitly consents, and can be withdrawn on request at any time.

What will be published

When the consented sample supports it, the following will be released under an open data licence:

The anonymised record set as CSV and JSON on GitHub and at docs.bms-ev.com/data/.
Methodology, including how SoH measurements from different tools were normalised and what was excluded.
Distribution of SoH at commissioning by battery family and calendar age.
Observed capacity change against equivalent full cycles for installations with at least 12 months of operation.
Fault-event rates by battery family and by inverter family.
Explicit statement of sample size, known biases and what the data does not support concluding.

Known limitations of this programme

Self-selection bias. Installations that fail early or are decommissioned are less likely to be reported than ones running well. Any published result carries this caveat.
Vendor-run study. BMS-EV sells the controller used in these installations. That is a structural conflict of interest and is stated on every release. Independent verification is actively sought — see below.
Heterogeneous SoH measurement. UDS diagnostics, ScanMyTesla, LeafSpy and OEM BMS reports do not all define SoH identically. The normalisation approach will be published alongside the data rather than hidden in it.
Small sample for now. The programme is new. Early releases will cover few installations and should be read as indicative, not conclusive.

How to contribute

If you run a second-life EV battery installation — whether or not it uses a BMS-EV controller — you can contribute a record through the integration report template on GitHub. The template includes an explicit consent checkbox. Installations using other controllers (Battery-Emulator, SimpBMS, Orion BMS, Batrium, REC) are welcome and will be recorded with the controller identified, because controller architecture is one of the variables worth tracking.

Researchers, installers or test laboratories interested in independent verification or joint publication can contact office@bms-ev.com.

References used in BMS-EV life projections

Until this programme produces its own data, the life projections quoted elsewhere in this documentation are derived from the following published sources combined with residential duty-cycle assumptions. They are estimates, not measurements of second-life stationary installations:

Preger et al. (2020), cycle-life characterisation of commercial lithium-ion cells across chemistries.
Recurrent Auto — EV fleet battery health reporting.
Geotab — EV battery degradation analysis from telematics fleet data.

Last updated: 2026-09-19

Status: Methodology published; dataset not yet released

Technical author: BMS-EV engineering team (Clima Boost Jakub Lipiński, Poland)

Reviewer: Jakub Lipiński, founder/lead engineer BMS-EV

Revision: 2026-09-19 — page created with data schema, anonymisation policy and declared limitations

Related: Case Studies · Compatibility Matrix · Public datasets