Run cufemlab against one of your motors.
cufemlab is self-serve at /app, and also available as an assisted pilot: you send a scoped 2-D PMSM case, we run the validated 2-D cogging/torque workflow on a GPU node, and you receive the waveform plus the FEMM/GetDP validation card and a signed report — all within a fixed scope and timeline. The self-serve app is live at /app — sign in to run these workflows yourself.
Three engagement models.
All tiers are by request. cufemlab is proprietary software — there is no public download. Pilots are scoped, priced individually, and operated under NDA on request.
Read-only access to documentation, sample notebooks, and the frozen benchmark report. For teams comparing methods and deciding whether to fund a pilot.
- Full documentation site
- Sample notebooks (download)
- Frozen benchmark report
- Email Q&A within 5 business days
We run the validated 2-D PMSM cogging/torque workflow on a GPU node we provision, ingest one scoped motor cross-section, and deliver a signed report with the cogging waveform and the FEMM/GetDP validation card.
- GPU node (RTX-class)
- Validated 2-D no-load cogging/torque run
- Independent FEMM/GetDP cross-check
- Signed report (PDF + JSON + Markdown)
- SHA-256 provenance on every output
- Engineering review call (60 min)
- 30-day exclusivity on findings
- Typical timeline: 2–4 weeks
Multi-motor programme, custom benchmark suite, on-prem deployment, or licensed integration into your CAD/PLM pipeline.
- Multi-month engagement
- Custom benchmarks & materials
- On-prem or cloud deployment
- Joint roadmap input on v0.4
- NDA available
- Dedicated engineering contact
What's in, what's out.
In scope (validated today)
- One motor topology (PMSM — radial-flux 2-D plane).
- 2-D linear-µ magnetostatics; airgap flux.
- No-load cogging/torque (Arkkio + virtual-work).
- Independent FEMM/GetDP cross-check and per-run validation card on the moving-band cogging workflow.
- Deterministic, HMAC-signed evidence report.
- One engineering review call (60 minutes) covering results and limitations.
Out of scope / roadmap (not validated)
- Operating-point torque — available in the app as an estimate (not the cross-checked pilot reference).
- Iron-loss estimate available in the app; full nonlinear-saturation core-loss — roadmap.
- Thermal and mechanical analysis — roadmap.
- NOT provided: a validated 3-D torque prediction. The 3-D lane is field/visualization only — it computes no torque. GPU-accelerated 3-D field analysis is available on compatible GPU workers, and runtime metadata reports the actual execution backend and fallback state, but that is a field solve, not a torque result.
- ASME/IEC certification (not what cufemlab does).
- Production sign-off (use ANSYS / Maxwell / Star-CCM+).
- No speed or accuracy advantage over a commercial tool is claimed (none measured).
Validated reference (the one canonical fact): On the frozen 4-pole/12-slot, 2 mm-airgap PMSM reference, cufem 2D matched GetDP at −0.64% amplitude full-gap with waveform Pearson r=0.9941, and FEMM at +5.30% amplitude-only, inside pre-declared ±20% amplitude and r≥0.95 thresholds. The 2-D no-load cogging cross-check (cufem = FEMM = GetDP, ~0.048 Nm) is the validated claim.
What you get at pilot close.
| Artifact | What it carries |
|---|---|
| Signed PDF report | Full verdict trail, cited references, declared tolerance, measured error per check, SHA-256 provenance hash. |
| JSON result file | Machine-readable verdict record; suitable for CI ingestion. |
| Markdown summary | Drop-in for pull-requests, reviews, or internal docs. |
| Notebook bundle | The exact notebook used to produce the run — re-runnable on your end. |
| Review call (60 min) | Walk-through of results, candid limitations discussion, v0.4 roadmap alignment. |
From brief to signed report.
Brief (week 0)
You send a one-page brief: motor family, what you want measured, available CAD/geometry, comparison data, timeline. We reply within 5 business days with a fixed scope, price, and timeline.
Run (weeks 1–3)
We run the validated 2-D cogging/torque workflow on a GPU node, ingest your scoped cross-section, and iterate against any internal sanity checks you supply.
Deliver (week 3 or 4)
You receive the signed report, the notebook bundle, and a 60-minute review call. Findings are exclusive to you for 30 days.
Honest about the cost.
cufemlab pilots are quoted per engagement. The Sandbox tier is free for qualified labs and academic groups; the Pilot tier carries a fixed fee covering engineering time + GPU node provisioning; Partner programmes are scoped over multiple months. Email info@secrotec.nl for an indicative number for your scope.