The orange arrow must point at the nose, along the long axis.

A 3D source-panel potential-flow solver with empirical viscous and base-drag corrections, Barrowman stability, a vortex-lattice lift cross-check, NACA flutter analysis, and an optional 3-DOF rigid-body trajectory. Everything runs locally in your browser. Inverted STL normals are detected and corrected automatically, since they silently invert the entire pressure field.

• Screening tool, not certification. Use it to rank your own design iterations, not to validate a flight.
• No lift from circulation in the CFD solve itself. Source panels have no Kutta condition, so the surface pressure/Cp view carries no fin lift. Fin lift instead comes from two independent methods computed separately: Barrowman's closed-form theory (drives the stability score) and a vortex-lattice method solved directly from the fin's planform (a numerical cross-check). The two can disagree by roughly 2× for low-aspect-ratio fins — validated against classical Jones slender-wing theory, this reflects a real difference between semi-empirical (Barrowman) and potential-flow (VLM) theory, not a computation error.
• 3-DOF trajectory is a pitch-plane (longitudinal) rigid-body simulation — wind, launch angle, aerodynamic normal force at the actual CP, and strip-theory pitch damping — not full 3D 6-DOF. Roll and out-of-plane motion aren't modeled, since they're not dynamically significant for a symmetric finned vehicle with no fin cant. Apogee, max velocity, and rail-exit velocity are validated. Max angle of attack and the pitch-oscillation trend are experimental: development testing found unresolved growth in pitch rate during long, decelerating coast phases even for statically stable configurations. This may be a real effect — simplified strip-theory damping models are known to omit terms (like moment due to the rate of change of angle of attack) that matter when dynamic pressure is changing quickly — or a residual modeling issue; it wasn't possible to fully distinguish the two before shipping. Treat these two outputs as informative, not authoritative, and use the static margin (caliber) as the primary stability check.
• Compressibility is a Prandtl-Glauert correction, valid to roughly Mach 0.7. Above that shocks and wave drag dominate and results are not meaningful.
• Base drag uses the Hoerner correlation assuming a blunt aft end and no boat-tail geometry detection.
• Flutter is the classic NACA TN-4197 estimate. It assumes a flat trapezoidal fin and ignores tip-to-tip layup, tab-to-fin bonding, and mounting stiffness — all of which matter in practice.
• Motor data uses class-midpoint impulse and typical burn times. Look up your actual motor.
• Mesh quality matters far less than it looks. Wetted area, stability, flutter, and drag are all reconstructed from the outer silhouette rather than summed from triangles, so holes, inverted normals, duplicated faces, and interior geometry barely move them. Against deliberately damaged test meshes, centre of pressure stayed within 0.3% and flutter within 0.0%. Only the pressure field genuinely needs a closed surface.
• Automatic repairs. Duplicate faces are removed, globally or partially inverted normals are corrected per-panel against the local radial direction, and interior geometry is culled before solving. The mesh health panel reports what was found and fixed.
• Cp values are indicative, not quantitative. Validated against the analytic sphere solution, the solver reproduces the pressure pattern and its extremes but under-predicts the stagnation peak by roughly half — a known limit of flat panels with point sources. Use the colours to see where pressure is high or low, not to read exact values. Reported drag does not depend on this, since friction and base terms dominate and are empirical.

Cross-check anything you intend to fly against OpenRocket or RockSim, and have it reviewed at the RSO table.