Open-Core · OpenFOAM-Powered · Local-First

Describe your flow problem in plain language.
Get a production-ready CFD pipeline.

SimuSpace translates natural-language descriptions into fully configured OpenFOAM workflows — geometry, mesh, physics, boundary conditions, and solver parameters — with confidence scores and explicit assumptions. No CFD expertise required to start.

simuspace — new simulation
Describe your problem:
"Simulate turbulent airflow over a 2D NACA 0012 airfoil at 5° angle of attack, Re = 3×10⁶. I need lift and drag coefficients and the separation point location."
Configuration generated (confidence: 94%)
SolversimpleFoam
TurbulencekOmegaSST
Mesh cells~1.2M
Est. runtime~45 min
BCs4 patches
Cores8
How it works

From problem description to solver-ready config

01

Describe in plain language

Write your fluid dynamics problem as you would explain it to a colleague. Include geometry, fluid, flow conditions, and what you want to measure.

02

AI generates CFD pipeline

The AI layer maps your description to a complete OpenFOAM configuration — solver choice, turbulence model, mesh parameters, and boundary conditions — with explicit confidence scores.

03

Review, refine, and run

Inspect every parameter, understand the AI's assumptions, override anything you disagree with, then export directly to OpenFOAM or run via the built-in solver interface.

Core features

Built for researchers who know the physics

Physically valid configurations

Every generated config is checked against OpenFOAM best practices. The AI flags assumptions, warns about underspecified problems, and never silently produces invalid setups.

CAD data stays local

Your geometry files never leave your machine. The AI configuration layer is the only part that touches external compute. A hard privacy guarantee — no exceptions.

Research-grade solver core

Built on OpenFOAM — reproducible, auditable, and community-maintained. No black-box cloud solver. Every result can be independently verified and published.

Version-controlled configs

Every simulation configuration is a structured JSON artifact. Diff runs, track changes, and reproduce any prior result — built for reproducibility from day one.

Turbulence model selection

The AI selects between laminar, k-ω SST, k-ε, Spalart-Allmaras, and others based on your Reynolds number, geometry, and flow regime — with an explanation for every choice.

Confidence scoring

Every configuration comes with a 0–100% confidence score and explicit list of assumptions. Low confidence triggers targeted clarification questions, not silent failures.

Start your first simulation

Free for individual researchers. No installation required for the configuration layer. Bring your own OpenFOAM runtime.

Open Workbench Free