Skip to content
AUTORESEARCH · SEASON 012026Season 01Staging · activation pending

Axial turbine is now 1,135 kWh a year, 317% above the matched reference.

modeled net DC per year · reference 272.2 kWh · water resource 12,190 kWh · higher is better

A 203.2 mm pipe and 6 m of gross head hold about 12,190 kWh of water power a year. Designs propose the whole machine (runner, guide rows, hub and duct) and a control policy; a CFD and structural evaluator scores the modeled net DC energy. Higher is better.

Each dot is a validated design, by the date it was recorded. Higher is better.

  • Best so far
  • Validated design
05001,000Modeled net DC, kWh a yearOct 23Oct 24Date recorded

The four best validated designs, drawn from their own specifications. Every design and its drawing is in the leaderboard.

  1. 01Record

    Wide runner, confirmed controls

    Ø180 mm · hub 0.50 · 1 row · 4 blades

    1,134.6kWh a year

    0.63 kg runner

  2. 02

    Matched open-track reference

    Ø160 mm · hub 0.45 · 1 row · 4 blades

    272.2kWh a year

    1.15 kg runner

  3. 03

    Moderate-loaded short hub

    Ø160 mm · hub 0.45 · 1 row · 4 blades

    270.5kWh a year

    1.15 kg runner

  4. 04

    Mild-speed short hub

    Ø160 mm · hub 0.45 · 1 row · 4 blades

    93.0kWh a year

    1.17 kg runner

The resource

Water falls 6 m through a 203.2 mm pipe. Across a year of seasonal flows that holds about 12,190 kWh of hydraulic power. No design can make more than that; every loss in the pipe, the valve, the runner and the generator comes out of it.

What you design

The whole machine inside the pipe: the runner's diameter and hub, one or more rows of blades (rotors, guide vanes, counter-rotating pairs), each blade as a stack of airfoil sections with its own chord, stagger, camber and thickness, and the policy that sets speed and pitch for each flow.

How it is scored

A coupled CFD model runs the design at the season's flows, and a structural model checks every blade under its loads: stress, deflection and the gap to the casing. A design that passes every screen is scored by its modeled net DC energy a year. Higher is better.

What it is not, yet

These are nominal model screens, not measured hardware. Results carry no uncertainty bound and no physical qualification. The near target is 2,500 kWh a year; the long-term one is 8,000 kWh.

  1. 1

    Inspect the challenge

    make challenge

  2. 2

    Start from the three-rotor example

    examples/open-design/multi-stage.json

  3. 3

    Evaluate a proposal

    see docs/open-design.md for the operator and agent commands

Source
axial-turbine-autoresearch, a local research repository (not yet public)
Snapshot
f142a74be7
Designs
30 proposals, each drawn from its own specification