orbitherm

Thermal digital twin for orbital compute — size the radiator, find the cooling wall.

In vacuum, heat leaves only by radiation (q = εσT⁴). A space data center must radiate away every watt it burns, plus what it absorbs from the Sun, Earth albedo, and Earth IR. Radiative capacity grows only as T⁴, so radiator area — not power — is the binding constraint on dense orbital compute. orbitherm sizes that radiator and gives a go/no-go verdict from first-order physics, with no heavy dependencies.

Install

pip install orbitherm

Quick start

import orbitherm as ot

twin = ot.ThermalTwin(
    power_w=100_000,                                  # 100 kW compute payload
    radiator=ot.Radiator(area_m2=200, orientation="deep_space"),
    orbit=ot.Orbit(altitude_km=550),
)

r = twin.evaluate()
print(r.verdict, f"{r.radiator_temp_c:.0f} C")        # -> viable 48 C
print(r.message)

# How big must a 1 MW data-center radiator be to hold 50 C?
print(ot.size_radiator(1_000_000))
# -> {'area_m2': ~1940, 'mass_kg': ~7760, ...}

What it models

• Steady-state radiator temperature — Stefan-Boltzmann rejection balanced against compute dissipation plus absorbed solar / albedo / Earth-IR loads.
• Required radiator area to hold a target temperature at a given power.
• Mass and launch cost (today's ~$1000/kg vs. a ~$200/kg breakeven target).
• Transient temperature over one orbit, including the eclipse swing, using radiator thermal mass.
• Orbit geometry — period and eclipse fraction from altitude.

Radiator orientation matters: deep_space (best), nadir, or sun_facing (worst case) change the parasitic environmental load.

API

Scope

First-order feasibility and sizing tool — good for go/no-go decisions, architecture trade studies, and pitch analysis. It is not a substitute for detailed CAD/finite-element thermal analysis. Assumes an isothermal radiator, a fixed radiator→junction temperature rise, and worst-case (β=0) eclipse geometry.

License

Apache-2.0