End-to-end digital twin for battery electrode manufacturing. SPH-based slurry rheology, wet-film & drying dynamics, and mechanical compaction physics — across all three stages of electrode production. Real-time process control built on top.
In pilot deployment with a tier-1 gigafactory. Production rollout planned for Q4 2026.
Mixing, coating, calendering — every step has a multiphysics signature. We model each stage with the right Physics-AI method, then chain them into a single digital twin of the line.
Four steps. The twin plugs into your existing line and your existing CMS — no new HPC, no new operators required.
Ingest line CAD, raw-material specs, and sensor telemetry from mixers, coating heads, and calenders.
SPH/DEM for mixing & granular flow, drying PINNs for coating, FEM-coupled ML for calendering compaction.
Twin runs at 5 Hz on the line PLC bus. SCADA-compatible. Predictions exposed via REST and a Python SDK.
Real-time recommendations: impeller speed, line speed, drying-zone temps, roll-gap setpoints — physics-grounded.
At full 3D + time, electrode-mixing physics is too expensive for pure PINNs. We use physics-guided ML surrogates — with constraints embedded in architecture and loss — so the model is fast enough to run line-rate and faithful to the physics underneath.
Full PINN training on 3D transient electrode flow is currently intractable at production scale. We get the same physical consistency by embedding mass / momentum / energy constraints directly into the architecture of a faster surrogate model — and we validate against high-fidelity SPH/DEM ground truth.
WASP & decarbonization decision intelligence for maritime. FuelEU & EU ETS compliance built in. Visit opsimnav.com ↗
/ Thermal · In developmentCross-sector thermal digital twin. Battery packs, heat exchangers, process thermal systems — one twin, three regimes.