Drone propeller by AI

Heliciel MCP: drone propeller designed by an AI

What you get: thrust per propeller, diameter, RPM, chord and pitch distribution, exportable 3D geometry — computed by Heliciel, not estimated by the model.

→ About the Heliciel MCP connector
→ Install in Claude
→ Get an MCP pass
→ Drone and multirotor propellers (article)
→ Version française

helice heliciel Drone propeller design with ChatGPT or Claude

With the Heliciel MCP connector, your AI assistant sizes the propellers of a drone or multirotor from the mass, the number of motors and the thrust margin you want. It does not guess a "common" diameter and pitch: it sets the thrust target, asks you for the available diameter, then has the blade computed by the Heliciel software (BEM, Blade Element Momentum) at your motors' RPM.

The problem: a drone propeller is computed in hover

A multirotor propeller spends most of its time at zero or low forward speed: induced velocity dominates, root incidence angles are high and efficiency depends strongly on the twist law. Rules of thumb (pitch ≈ 0.4 × diameter, "this motor takes that prop") give neither the real thrust nor the absorbed power. A language model asked alone repeats those rules of thumb; it does not compute. The connector replaces the rule with the computation. Background: drone, VTOL and multirotor propellers.

Real dialogue: 8 kg drone, 4 motors

Real session in the ChatGPT iPhone app (from the demo video), Heliciel connector active.

"I have an 8 kg drone, 4 motors, 150 % thrust margin, motors at 5000 rpm. Size the propellers."

What the assistant does. It sets the target: 8 kg × 9.81 m/s² × 1.5 / 4 motors = 29.4 N per propeller. Then it asks for the maximum diameter — it did not receive it and Heliciel refuses to invent it. First design returned by Heliciel: Ø 376 mm, 3 blades, 5,000 rpm, propulsion_air project in hover.

"Max diameter 400 mm."

Updated design by Heliciel: Ø 400 mm, 3 blades, hub 44 mm, 5,000 rpm → thrust 65.0 N per propeller, 260 N for the four motors. The assistant compares with the 29.4 N target and states the actual margin obtained (well above the requested 150 %) instead of hiding it. Heliciel reports 0 errors and 0 warnings on this design.
What the AI refused to invent. The diameter. In a conversation without the connector, the model would have picked a "typical" value. Here Heliciel does not start the computation without the data: the diameter is a design lever that belongs to you (frame size, clearance between propellers).
3-blade drone propeller in the Heliciel 3D viewer on ChatGPT iPhone: front, side, top views, STL, OBJ, IGS, cavitation buttons, 5000 rpmThe Heliciel 3D viewer in ChatGPT iOS: the 3-blade propeller rotates under your finger; Hub, Sections, STL, OBJ, IGS, Alerts buttons within thumb reach. 5,000 rpm, forward speed ≈ 0 (hover).
Real screenshot of the Heliciel software returned in ChatGPT mobile for the drone propeller: interface, 3D blade, lift, thrust and power chartsThe visualiser tool returns Heliciel's real screen ("Computed by Heliciel — not AI-generated"): tabs, 3D blade, Cl/Cd charts, thrust and power.

What you can ask next

Limits specific to drone propellers

Connect

  1. Get a Heliciel MCP pass (per day) on mecaflux.com.
  2. In Claude: install from the connectors directory. Elsewhere: add an MCP connector at https://mcp.heliciel.com/mcp (per-client details).
  3. Authorize with the pass key, then write your request as above.

Frequently asked questions

Can the assistant choose the diameter for me?
No. It can help you reason (clearance, frame size), but Heliciel does not accept a diameter you did not give.

Does the result hold in forward flight?
The design above is in hover. Then ask for an off-design analysis at the forward speed you want: same propeller, new operating point.

Can I reproduce these figures?
Yes: same prompt, same project type, same diameter and RPM give the same result. It is a deterministic computation, not a generation.

Related pages: Heliciel MCP (overview) · aircraft propeller by AI · wind turbine by AI