Joule
An integrated instrument for battery survival in the hardest power environments — the cold that plates lithium, the heat that ignites a cell, the thermal hammer of a lunar duty cycle. From a single engine, Joule answers the core questions of cell survival from the electrode's own pore-microstructure: how fast it can charge in the cold before lithium plates; the charge rate at which an internal hot-spot self-ignites; how many cycles it lasts under a thermal duty cycle and how it fails; and the densest electrode that survives all of it. The same micro-CT returns the same assessment, reproducible and auditable, in seconds where conventional case-by-case testing takes weeks to months.
The observables are classical — effective ion transport and the tortuosity factor, the Frank–Kamenetskii ignition threshold, Arrhenius degradation, the percolation transition. The method is published and dedicated to the public under the MIT license (DOI 10.5281/zenodo.20866741). The operational engine is not distributed.
What it answers
One battery specification — a segmented micro-CT of the electrode — in; deterministic answers out.
| Question | Joule returns |
|---|---|
| COLD LIMIT | the safe fast-charge envelope versus temperature, and the pore where lithium plates first |
| HOT LIMIT | the charge rate at which an internal hot-spot crosses the ignition threshold — and the wall no external cooling can pass |
| MISSION LIFE | the cycles to end-of-life under a thermal duty cycle, and the dominant failure mode (cold plating versus hot growth) |
| SAFE DENSITY | the densest electrode that survives both limits — energy density against the safe operating window |
| CERTIFICATE | a deterministic, bit-reproducible CONTAINED / MARGINAL / ESCAPED verdict, with provenance, that never exposes the geometry |
All deterministic: the same micro-CT returns the same assessment, reproducible from the classical definitions — no random sampling, no fitted model, no per-cell tuning.
Demonstrated — reproducible against the open standard
Joule's transport observable is the classical tortuosity factor — the same quantity the open-source reference (TauFactor, Imperial College) computes from tomography. The claim is reproducible: on a public electrode micro-CT, against the public tool, anyone can verify it. Joule returns the same number, and returns it faster.
| Test | Result |
|---|---|
| Tortuosity factor — real 256³ NMC electrode micro-CT | matches the open standard to 0.121% |
| Exact analytical geometry (straight channels) | machine-precise — τ = 0.99999 vs the closed form 1 |
| Versus the Bruggeman textbook correlation | 1.4× more tortuous (the optimism a datasheet hides) |
| Same volume, same machine | 3.47× faster, at a tighter tolerance |
| Full-pack scale — one desktop computer | certifies a 3.23-billion-element electrode |
Validated against the open-source standard on its own public 256³ electrode micro-CT — data Joule did not generate. Reproducible: the comparison runs on public data against a public tool. Method, fixture, and configuration: 10.5281/zenodo.20866741.
View a sample battery analysis — a lunar cell's life, to thermal runaway →
Scope — stated plainly
- The observables are classical; the method is what is new. Joule proposes no new electrochemistry — the tortuosity factor, the ignition threshold, Arrhenius degradation, and the percolation ceiling are all standard. The contribution is the exact, deterministic, integrated delivery of all of them from microstructure alone, in seconds rather than weeks.
- The transport observable is externally validated; the aging thresholds carry calibration. The effective transport (tortuosity / effective diffusivity) reproduces the open-source standard to machine and benchmark precision; the absolute ignition and degradation scales are calibrated to reference data and stated per engagement.
- A public-benchmark and canonical demonstration, not a population study. Validated on a public electrode micro-CT and exact analytical geometries; a broad cell-catalog validation is the next step.
- The engine is patent-pending and undisclosed. The observables reproduce from the classical definitions; the speed and exactness come from a sealed engine — the subject of U.S. Provisional Patent Application No. 64/098,689 (filed June 25, 2026), held as a trade secret and not published.
Cell and pack designers, extreme-environment and aerospace battery programs, cold-climate and fast-charge fleets, and underwriters of battery risk: email with your cell and operating envelope, and a scoped evaluation follows — a reproducible certificate for a real cell, not a stochastic estimate.