Grounded in research.

Abstract

Argues that vehicle-level battery State of Health should be standardized around two transparent metrics — energy-based (usable energy now ÷ original usable energy) and capacity-based (usable charge now ÷ original) — rather than ad-hoc range proxies. Proposes a reproducible, scalable measurement procedure that leverages ordinary on-board charging data, so SoH can be determined in the field without lab equipment. Directly motivates measuring SoH from charge sessions.

Why this matters for your car

Motivates showing you an energy-based SoH and a charge-session-measured capacity instead of relying only on rated range. Our estimate is INSPIRED BY this paper's on-board-charging idea but does not reproduce its controlled procedure (fixed voltage windows, low power, temperature control, high-rate data); we use ordinary daily charge-session data, so we label ours an estimate.

Abstract

Large-scale real-world EV battery degradation analysis. Analyzed telematics from 22,700+ electric vehicles across 21 models in commercial and private fleets between 2018-2024. Found average annual degradation of 2.3% across all models, with no evidence of premature cliff drops below 70% capacity in the first decade. Geotab's methodology compares each vehicle's reported full-charge range against EPA spec, normalized for ambient temperature and odometer.

Why this matters for your car

Your battery health score is benchmarked against this 22,700-vehicle dataset. The percentile shown on the fleet-comparison card tells you where your car sits among real fleet data, not Tesla marketing claims.

Abstract

The EU Euro 7 regulation extends type-approval to EV battery durability, adopting the UN GTR 22 structure with slightly stricter floors for passenger cars: ≥80% of original capacity at 5 years or 100,000 km, and ≥72% at 8 years or 160,000 km (vans: 75% / 67%). It also mandates that owners get on-board access to up-to-date battery-health information. It enters into application from late 2026.

Why this matters for your car

Euro 7 is the EU's adoption of the durability corridor. We anchor our health corridor to the more conservative UN GTR 22 floor (70% at 8 years) and note that Euro 7 holds passenger cars to a tighter 72% at the same point.

Abstract

Aggregated range-versus-speed measurements from controlled-track tests of all major EVs. Drag scales with the square of velocity, so range scales inversely with v² above ~50 mph. Empirical fit: a Tesla losing 18-22% of EPA range at a sustained 75 mph vs the EPA 60 mph average. Wind, rolling resistance, and accessory load are accounted for separately; the aerodynamic component dominates above 55 mph.

Why this matters for your car

The speed-range table on this page projects how far you'll actually go at 60/65/70/75/80 mph based on your car's specific EPA range — useful for trip planning, especially with kids/cargo loaded.

Abstract

Synthesis of manufacturer and research-typical design cycle-life figures used for headroom estimates: LFP cells are rated for roughly 3,000+ equivalent full cycles to 80% capacity, NCA/NMC 18650/2170 cells roughly 1,500, and the newer 4680 cell is treated as an NCA-class proxy (~1,500) pending independent long-cycle data. Tesla's own guidance instructs LFP-pack owners to charge to 100% regularly (at least weekly) for accurate state-of-charge calibration, reflecting LFP's tolerance of high SoC.

Why this matters for your car

Your cycle-headroom percentage divides your estimated equivalent full cycles by the chemistry's design rating from this synthesis. The 4680 rating is explicitly labeled an estimate. The LFP tuner recommendation to charge to 100% periodically comes from Tesla's calibration guidance, not generic advice.

Abstract

A comprehensive review of lithium-ion SoH estimation from cell to pack level, organizing the field around resistance-, capacity-, and energy-based SoH indices. Confirms coulomb counting — integrating charge throughput against the state-of-charge change during a charge/discharge — as a high-accuracy, direct capacity-measurement method, while noting its dependence on accurate SoC and sufficient charge windows.

Why this matters for your car

Backs our measured-capacity card. We estimate usable capacity as charging ENERGY added (kWh) ÷ state-of-charge rise across a charge session — an energy-counting proxy in the same partial-charge capacity-estimation family this review surveys. (Strict coulomb counting integrates charge in Ah; we use energy, which is closely related.) The key point it supports: such estimates are independent of rated range.

Abstract

Recurrent Auto aggregated battery health reports from 15,000+ Tesla owners between 2019-2023, segmented by trim, chemistry (LFP vs NCA/NMC), region, and charging behavior. Key findings: LFP packs in Standard Range Model 3/Y show essentially flat degradation through 50,000 mi when charged to 100% routinely. NCA Long Range and Performance trims show ~5-7% loss by 30,000 mi if frequently DC fast-charged, ~2-3% if Level 2 home-charged.

Why this matters for your car

The chemistry-aware charging recommendations (LFP: 100% daily OK; NCA: keep below 80% daily) come from Recurrent's per-chemistry breakdown, not generic Tesla advice.

Abstract

The first internationally-harmonized minimum performance requirements for the durability of EV traction batteries. For light-duty vehicles it sets a State of Certified Energy (SOCE) floor of ≥80% of the original usable energy from start of life to 5 years or 100,000 km (whichever comes first), and ≥70% up to 8 years or 160,000 km. SOCE is an energy-based metric (usable energy now ÷ certified original energy), measured via standardized on-board monitoring. It is being progressively adopted into EU (Euro 7), North American (EPA/CARB), and Japanese regulation.

Why this matters for your car

Your energy-based State of Health (SOH_E) is benchmarked against this corridor: we show whether your pack's energy capacity is above the regulatory minimum for your car's exact age and mileage. It is a regulatory reference point, not a Tesla warranty test.

Abstract

Companion cycle-aging study to Naumann 2018, on the same LFP/graphite cell. Cycle capacity loss scales as a power law of charge throughput (equivalent full cycles) with an exponent below 1 — fade per cycle slows as throughput accumulates. Depth-of-discharge and C-rate are secondary factors. Confirms LFP's very high cycle life (thousands of equivalent full cycles to 80% capacity).

Why this matters for your car

Your equivalent-full-cycle model uses this power-law cycle exponent (z≈0.55 for LFP) so that the cycle-fade contribution flattens out the way real LFP cells do, rather than growing linearly. It underpins the 'you have thousands of cycles of headroom' verdict for LFP cars.

Abstract

AAA tested five EVs — BMW i3s, Chevrolet Bolt, Nissan Leaf, Tesla Model S 75D, and VW e-Golf (the Model S 75D used a resistive cabin heater) — in 20°F (-7°C) weather using both heater and pre-conditioning. Found ~41% average range loss with heater on vs EPA, ~12% with pre-conditioning. Range loss varied by vehicle and was not lowest for the Teslas. AAA's per-degree-F derate curve is the basis for most third-party cold-weather estimators including Tesla's own in-car projection.

Why this matters for your car

The cold-weather impact card applies AAA's curve to your ambient temperature (when below 60°F). What looks like sudden battery degradation in winter is usually heater + cabin demand and is fully recoverable.

Abstract

Long-term calendar-aging study of a commercial LFP/graphite cell across temperatures and storage SoC levels. Capacity loss follows a square-root-of-time law; the rate depends on temperature via Arrhenius (LFP activation energy ~17 kJ/mol — markedly lower than NCA/NMC, so LFP is less heat-sensitive) and increases with storage SoC. Establishes that LFP calendar aging is slow and well-behaved, supporting high-SoC daily use.

Why this matters for your car

For your LFP pack the calendar model uses Naumann's lower activation energy and √-time exponent, which correctly predicts that high charge levels cost an LFP battery far less calendar life than they would an NCA pack — the basis for the 'charge to 100% is fine' guidance.

Abstract

Controlled long-term storage study tracking capacity loss in NMC and NCA Li-ion cells held at various state-of-charge levels and temperatures over 12-24 months. Showed clear monotonic relationship: cells stored at 100% SoC degraded ~2.5x faster than those held at 50%, even at room temperature. Combined SoC and temperature stress (e.g. 100% + 40°C) accelerated loss to 5x baseline. Data underpins Tesla's own daily-charge-limit recommendations.

Why this matters for your car

Your charge-limit card scores how much extra calendar aging your daily-usage habits cost. Setting a 90% limit on an NCA car instead of 100% can preserve ~15% more lifetime capacity over 5 years.

Abstract

Holistic semi-empirical aging model for NMC/NCA 18650 cells separating calendar from cycle aging. Calendar capacity loss follows a power law of time with an exponent of ≈0.75 (t^0.75), whose rate constant depends on temperature (Arrhenius, an NMC calendar activation energy on the order of ~50-60 kJ/mol) and on storage voltage/SoC (a near-linear-to-quadratic voltage term — higher SoC accelerates loss). Cycle aging scales with charge throughput and depth-of-discharge. The model is widely used to attribute observed fade to its calendar vs cycle components.

Why this matters for your car

Your calendar-vs-cycle split uses the Schmalstieg functional form (Arrhenius temperature term × SoC term × t^0.75 time law) to estimate what fraction of your pack's fade comes from sitting at a high charge level versus from driving miles. For NCA/NMC/4680 packs this is the dominant aging family. (LFP uses Naumann's slower √-time law instead.)

Abstract

NREL study applying an additive, separately-calibrated semi-empirical battery life model to project PHEV graphite/NCA pack life across U.S. geographies and drive cycles. Calendar and cycle fade are modeled as ADDITIVE terms: a calendar component (Arrhenius temperature dependence plus an SoC/anode-potential dependence, evolving with the square root of time) added to a cycle component proportional to charge throughput. The framework is chemistry-agnostic in FORM; each term's coefficients are calibrated from lab data for the specific cell.

Why this matters for your car

We borrow NREL's additive STRUCTURE — a calendar term (temperature × SoC × time) added to a cycle term (throughput) — as the functional form for our split. It does not validate our specific sparse-data attribution: we reference-normalize the two terms and anchor their sum to your car's measured fade, which yields a directional estimate, not a calibrated measurement.

Abstract

Foundational cycle-life model for graphite/LFP cells fit to a large test matrix of temperature, depth-of-discharge, and C-rate. Capacity loss follows a power law of charge throughput (Ah) with an exponent of ~0.55 — sub-linear, so per-cycle fade decreases as cumulative throughput grows. Temperature enters via Arrhenius. The power-law throughput exponent is widely reused as the canonical cycle-aging form across chemistries.

Why this matters for your car

The equivalent-full-cycle math uses Wang's z≈0.55 throughput power law so that your cycle wear is estimated from lifetime miles → energy → equivalent full cycles, with the realistic sub-linear slowdown. This is also what bounds the design cycle-life ratings used to compute your cycle headroom.

Abstract

Foundational equation in physical chemistry: reaction rates roughly double for every 10°C rise in temperature. For lithium-ion battery cells, this manifests as accelerated calendar aging — the side reactions at the SEI layer that consume lithium and electrolyte over time speed up exponentially with heat. Modern Li-ion battery aging models (e.g. Smith & Wang 2017, Schmalstieg et al. 2014) all build on this thermal kinetics baseline.

Why this matters for your car

Your climate-impact card uses Arrhenius doubling rule on the recorded ambient temperatures from your area to estimate how much faster your battery ages versus a car parked in San Diego (~25°C / 77°F annual avg).