● Beta OpSimNav — WASP & decarbonization decision intelligence — is now live. Visit OpSimNav →
OpSimTech / Products / OpSimBat
OpSimTech · OpSimBat

OpSimBat

The Physics-AI Layer For
Battery Electrode Manufacturing.

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.

/ Status In development · Pilot 2026

In pilot deployment with a tier-1 gigafactory. Production rollout planned for Q4 2026.

Request a Demo See our method
−34%
Electrode defect rate · pilot fleet
+7.5%
Line uptime · post-deployment
3
Stages covered · mix · coat · cal.
5Hz
Real-time twin update rate
/ 01 — Pillars

One twin. Three stages.

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.

/ 01 Stage 01 · Slurry preparation

Mixing

SPH-based slurry mixing twin. Electrode paste rheology, particle dispersion, dead-zone detection. Real-time process monitoring catches mixing failures before they hit the coating head.

SPH DEM Rheology Dead-zone
1L → 10
Tank scales
±2.4%
Uniformity vs spec
SPH mixing · cathode slurry P01 · MIX
/ 02 Stage 02 · Thin-film deposition

Coating

Slot-die thin-film coating twin. Wet-film uniformity, drying front dynamics, edge effects, and defect-formation prediction. Catch streaks, pinholes, and edge bead issues before the line degrades quality.

PINNs · drying Physics-guided ML Slot-die Defect map
±3μm
Film thickness target
< 50ms
Defect-map latency
Slot-die · wet film + drying P02 · COAT DRYING UNWIND REWIND
/ 03 Stage 03 · Mechanical compaction

Calendering

Mechanical compaction twin. Porosity control, density uniformity, crack and delamination prediction. The twin predicts how each set of roll-gap conditions translates into cell-level cycle performance downstream.

FEM + ML DEM coupling Porosity Crack prediction
±0.5% porosity
Targeted control
+12%
Cycle life · downstream
Calendering rolls · porosity field P03 · CAL. IN OUT
/ 02 — How it works

From CAD & sensors to live process control.

Four steps. The twin plugs into your existing line and your existing CMS — no new HPC, no new operators required.

/ 01

Capture

Ingest line CAD, raw-material specs, and sensor telemetry from mixers, coating heads, and calenders.

/ 02

Model

SPH/DEM for mixing & granular flow, drying PINNs for coating, FEM-coupled ML for calendering compaction.

/ 03

Deploy

Twin runs at 5 Hz on the line PLC bus. SCADA-compatible. Predictions exposed via REST and a Python SDK.

/ 04

Control

Real-time recommendations: impeller speed, line speed, drying-zone temps, roll-gap setpoints — physics-grounded.

/ 03 — Method

Why physics-guided surrogates for battery mfg.

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.

Honest about tractability.

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.

Primary method · Physics-Guided Surrogates
  • Mass & momentum conservation baked into network architecture
  • Loss includes physics-residual terms where tractable
  • SPH / DEM serve as offline validators & trainers
  • 5 Hz inference on commodity edge hardware
  • Uncertainty-quantified outputs — calibrated error bars
/ 04 — Sister products

Other OpSim products.

/ 05 — Get in touch

Ready for a
smarter electrode line?
Let's pilot together.

Request a Demo Email an engineer