Overview
spacecost is the transportation stage of economicspace, my asteroid-mining
profitability pipeline, split out as its own Python package because a launch price is useful to anyone costing
a mission. Every row carries an inline citation and the year its figure refers to, so a stale price is visible
rather than hidden. On top of the tables, the package derives what they imply: tank mass from propellant
volume, blended bipropellant performance, and the cost of delivering a kilogram to the Moon or Mars stage by
stage. It installs straight from GitHub and has a small command-line interface for ranking launchers and
propellants.
Tables
Concept and retired rows are kept and marked, so a study can say what it excluded rather than silently not having it.
| Table | Rows | Holds |
|---|---|---|
| Launch vehicles | 76 | $/kg to LEO, GTO and escape with a low/high band, payload, status, and who can buy it |
| Propellants | 41 | Vacuum Isp, density, price, storage class, derived tankage, thruster device |
| Delta-v segments | 33 | m/s and trip time per trajectory leg, 100 to 10,500 m/s |
| Operational costs | 45 | Per-year and per-kilogram mission costs with low/high bands |
| Storage systems | 20 | How a kilogram is held, across cargo, depot, propellant and energy |
| Environments | 23 | Solar flux, dark period, gravity and light time at each destination |
Key highlights
What it does that a spreadsheet of prices does not.
- Prices are the centre of a stated range. Where a source gives a range, the headline is its geometric centre, because launch prices are uncertain by factors. The Space Shuttle's $450M–$2.1B per flight centres at $972M, where a plain average would say $1.28B. Unflown vehicles' targets only ever count as the optimistic end.
- Flying is not the same as for sale. 28 of the 48 operational launchers fly only for somebody else, because they are sold out, government-only or export-restricted, and the rankings can filter to the ones a buyer can actually book.
- Tank mass is derived, not asserted. A tank scales with the volume it encloses, so it is sized from each propellant's density and storage class and anchored on real flight hardware. Hydrolox comes out at 9.7% of propellant mass, matching Centaur's measured ~9.7%.
- Deliveries are staged leg by leg. Chaining real stages to the lunar surface needs 4.99 kg in LEO per kilogram landed; collapsing the same trip into one delta-v figure would say 10.96. Stages that get thrown away are paid for, not just launched.
- The rocket equation is not taken literally. Solar sails have infinite Isp, which made every delta-v cost $0.00 and swept them to the top of every ranking, so they are excluded by default. Electric stages carry a 1.5× spiral penalty, and the adjusted delta-v is reported rather than hidden.
- The output is a contract. The six tables build byte-identical on every platform, because their derivations use only operations IEEE 754 requires to be correctly rounded. Continuous integration checks that on every push.
Links
The package, its tests and its changelog are public. Figures on this page are from spacecost 0.5.0 (data contract 1.17.0).