Prepared for Captura · partnership proposal · July 2026
Direct ocean capture, modeled
Your process is electrochemistry. So is our engine.
Captura pulls CO₂ from seawater with bipolar-membrane electrodialysis. MESSAI is an electrochemical-systems modeling platform. What we bring isn't a literature match — it's the ion-transport physics, membrane models, calibrated prediction, and research tooling that sit underneath the platform, and that map cleanly onto your stack.
A mechanistic ion-transport + carbonate/bicarbonate pH engine, a membrane parameter library, uncertainty-calibrated performance prediction, techno-economic modeling, and a paper-extraction → knowledge-graph stack. None of it cares whether the electrons come from a microbe or a rectifier.
- Our corpus is microbial (biotic) electrochemistry; your process is abiotic.
- Our priors are fit on microbial data and would need re-fitting on electrodialysis data before any prediction is trustworthy.
- Day-one value is the engine and physics modules — not turnkey numbers.
Our 1-D Nernst–Planck solver with exact carbonate/bicarbonate speciation models CO₂ flux vs. current density, alkalinity/DIC, and temperature — the acidification/degas step, directly.
Our catalog already types BPM / CEM / AEM by permselectivity, area-specific resistance, conductivity and durability — the levers behind your voltage-degradation and energy-per-ton claims.
Calibrated prediction and a TEA sandbox project energy intensity, capture rate, membrane life and $/ton across your 1t → 100t → 1,000t path — with honest confidence intervals.
Repoint our Nernst–Planck + buffer-chemistry solver from a biofilm to a membrane channel: predicted CO₂ flux and the proton-transport-limited "critical current density" regime.
Map ASR, permselectivity and degradation rate to stack voltage → energy intensity → $/ton. Compare candidate membranes on one consistent yardstick.
Hierarchical priors + conformal calibration emit a strict {value, unit, ci_low, ci_high, confidence, source} contract — error bars investors and DOE will ask for.
Model $/ton CO₂ against current density, membrane cost/life, electricity price, plant scale and infrastructure reuse — with the sensitivity view that shows which lever matters.
Point our extraction pipeline + "Hunter" research-scan at the DOC / mCDR / BPMED literature — a living, queryable KG of energy intensity, current density, fouling and MRV methods.
Encode your cause-graph — alkalinity → local pH → degas efficiency → energy/ton; current density → voltage → membrane life → capex — for internal reasoning and clear explanation.
Our R3F / Three.js stack renders a membrane-stack + flow-channel + degassing twin — useful as both an engineering tool and an investor-facing asset.
SI unit normalization and canonical parameters turn three pilot sites' heterogeneous logs into one comparable dataset, and formalize reproducible MRV protocols.
Here our biotic corpus earns its place: microbial electrosynthesis converts CO₂ into fuels and chemicals — a downstream reuse pathway for the CO₂ you capture.
Microbial desalination cells share your ion-exchange-membrane + acid/base architecture. Learnings on fouling, scaling and ion transport flow both directions.
We already model the exact chemistry you exploit
MESSAI ships a 1-D Nernst–Planck solver with diffusion + electromigration flux and exact carbonate / bicarbonate / phosphate buffer speciation, including the proton-transport-limited critical-current-density regime. It was written for biofilm pH gradients — but HCO₃⁻ → CO₂ under local acidification is your process. The boundary conditions move from an electrode surface to a membrane channel; the physics is the same.
Run extraction over ~150–200 DOC / mCDR / BPMED papers into a Captura-relevant parameter knowledge graph.
Move the Nernst–Planck model to a membrane channel and reproduce a published CO₂-flux-vs-current-density curve.
A small $/ton sensitivity model across current density, membrane life and electricity price.
Cheap to run, and it de-risks the whole thesis: it shows whether the physics-engine transfer holds before anyone commits to more.
Let's run the pilot.
MESSAI — electrochemical-systems modeling platform.