OpenTsiolkovsky

English | 日本語

!OpenTsiolkovsky

OpenTsiolkovsky is an open-source flight simulator for multi-stage rockets, from launch to impact. It runs as a command-line tool and as a web app that computes everything in the browser: open-tsiolkovsky.vercel.app.

What it does

• Flies sounding rockets and orbital launchers as a point mass (3DoF) with a commanded attitude program, over a rotating WGS84 Earth with J2 gravity, the U.S. Standard Atmosphere 1976 and wind.
• Models stages, engines (thrust and Isp as constants or tables), per-stage aerodynamics, jettisoned parts, a launch rail and thrust trim.
• Reports flight events (liftoff, Mach 1, max-Q, burnout, separation, apogee, impact) with orbital elements, and the impact points of the vehicle, spent stages and fairings.
• Runs Monte Carlo dispersion analysis and draws 1σ / 95 % / 99 % impact ellipses.
• Web app: 3D globe, charts and playback; sample vehicles (MOMO, S-520, Epsilon, H-IIA, H3, Falcon 9, Saturn V, …) built from published data; share links; Japanese and English.
• Can be driven by AI agents through JSON output and an MCP server.

Getting started

Web app

Open open-tsiolkovsky.vercel.app, pick a sample vehicle from the header and edit it. Results update as you type. The file menu saves and opens settings, exports CSV and KML, and copies a share link (the settings travel in the URL and are never sent to a server).

Command line

Download the archive for your machine from Releases (Windows x64, macOS on Apple Silicon, Linux x64 / Ubuntu 22.04 or later), unpack it and run openTsiolkovsky-cli from that folder; the sample vehicles are included. On macOS, if the binary is blocked, run xattr -d com.apple.quarantine openTsiolkovsky-cli once. To build from source instead, install Rust (stable) and run cargo build --release.

B=./openTsiolkovsky-cli                          # or ./target/release/openTsiolkovsky-cli
$B run examples/momo.json -o out/momo            # trajectory.csv, summary.json, bodies/*.csv
$B monte-carlo examples/SS-520-4/ss520_4.json -n 500 -o out/ss520
$B validate my_rocket.json                       # check a configuration without flying it

A configuration is a JSON file. Keys end in their unit, and anything that varies can be a constant, a table or a CSV file:

{
  "name": "minimal",
  "simulation": { "end_time_s": 300 },
  "launch": { "latitude_deg": 35, "longitude_deg": 139 },
  "attitude": { "azimuth_deg": 90, "elevation_deg": [[0, 90], [10, 90], [60, 70]] },
  "stages": [{
    "mass_initial_kg": 1000,
    "engine": { "thrust_vac_n": 30000, "isp_vac_s": 200, "burn_end_s": 20 },
    "aero": { "diameter_m": 0.5, "ca": 0.4 }
  }]
}

examples/ holds the web app's sample vehicles as configuration files: momo, s310, s520, ss520-5, epsilon, h2a202, h3-22s, falcon9, saturn-v, ariane5, soyuz-2.1a and long-march-3b (SS-520-4/ shows CSV tables). The web app reads and writes the same files.

AI agents

$B run examples/momo.json --json      # one JSON document on stdout
$B schema                             # JSON Schema of the configuration
$B share examples/momo.json           # link that opens the vehicle in the web app
claude mcp add opentsiolkovsky -- "$PWD/target/release/openTsiolkovsky-cli" mcp

See docs/ai.md for the JSON format, exit codes and MCP tools.

Documentation

The documents below are in Japanese.

• Configuration: every key, units and defaults
• Output files: trajectory.csv, summary.json, Monte Carlo files
• Physics models and numerics
• Trajectory optimization: single-stage MOMO and assumed multistage orbital payload optimization, native/WASI solver, browser forms, 3D results, saved runs and independent integration checks
• AI agents: JSON output, MCP server, share links
• Development: building, testing, WASM, project layout
• Web app internals

License

MIT