MISSION 01 / EARTH → ORBIT
A little closer to the sun.
A lot of heat to lose.
Deliver a data center to orbit. Harvest sunlight.
Find the limit between computing and cooling.
- 01 Launch
- 02 Ascent
- 03 Deploy
- 04 Compute
ROCKET → DATA CENTERPayload release
Scrub to inspect · replay with Watch deployment
ORBIT PRESETSCompare temperature & output ↓
Starting the 3D mission…
ROCKET DESIGN / PREDICT · LAUNCH · COMPARE
Give it a different chance.
Sets the engine flow limit for both stages. The booster burns at this setting; upper-stage guidance can throttle below it.
Scales both stages’ propellant loads. More fuel also means more launch mass.
Payload stays attached during ascent. This changes launch mass; the orbital thermal hardware settings remain independent.
Calculating flight…
Calculating…
Projected result for these settings. Orbit requires a bound trajectory with perigee at least 150 km in this demo. The target band is 450–550 km. Crossing 100 km alone does not count as orbit.
PROJECTED PERIGEE—PROJECTED APOGEE—END PROPELLANT / ATTACHED—MODELED FLIGHT TIME—
Perigee is the lowest altitude of the implied orbit; a negative value means the trajectory intersects Earth. Apogee is its highest altitude.
Blue dotted line: 100 km reference. Both traces use actual integrated motion and a shared time scale. Failed flights remain in the launch view; the thermal lab opens automatically only after orbital success.
LAUNCH LAB / CONFIGURABLE FLIGHT DYNAMICS
From force to flight.
On pad · 505 tonnes of fuel and oxidizer loaded.
Flight duration follows the configured rocket.
VELOCITY0.00 km/sATTACHED MASS542.5 tLOCAL GRAVITY9.82 m/s²DYNAMIC PRESSURE0.00 kPa
The 100 km Kármán line is a reference boundary. Air thins gradually; crossing it does not mean escaping Earth’s gravity or reaching orbit. Acceleration above is the change in speed along the flight direction, including gravity and drag.
Thrust equation, propellant tanks & model assumptions
F = ṁvₑ + (pₑ − pₐ)Aₑ
Engines off · launchpad supports the stack.
- ṁ · propellant mass flow
- Engine off
- vₑ · exhaust exit velocity
- —
- pₑ · nozzle exit pressure
- —
- pₐ · ambient pressure
- 101.325 kPa
- Aₑ · total nozzle exit area
- —
- Aerodynamic drag
- 0.0 kN
Thrust is a force (MN). Stack momentum is mass × velocity (MN·s). Momentum can decrease while speed rises because exhaust and detached stages carry mass away.
PROPELLANT / FUEL + OXIDIZER
First stage 410.0 t remaining
Upper stage 95.0 t remaining
The booster separates when its propellant is exhausted. Upper-stage reserves, when available, remain after cutoff. Momentum and mass refer to the attached stack before payload release.
This fictional two-stage vehicle integrates motion in a nonrotating 2D Earth frame using thrust, changing mass, inverse-square gravity, and drag. Engine flow is bounded by your thrust setting. The booster follows a fixed pitch schedule. Upper-stage injection raises apogee toward 500 km, followed by an engine-off coast and a prograde circularization burn. Pitch changes are limited to 2°/s; guidance commands outward or prograde thrust, with bounded flow and throttling. Orbital elements from position and velocity determine success. A 150 km perigee is a demo threshold, not a guarantee of orbit lifetime. Earth rotation, winds, structural loads, tank-volume constraints, gimbal limits and rotational inertia are omitted. Air pressure and density use a simple 8.5 km exponential scale height, not a high-altitude atmosphere forecast. The payload stays folded inside its fairing until fairing release at 110 km, then deploys its arrays and radiators after final cutoff. Vehicle sizes, low-altitude visual spacing, the 22-second deployment timing, and the transfer into another selected orbit are schematic. Telemetry retains the integrated physical altitude.
NASA — rocket thrust equation ↗ · NASA — orbit, perigee & apogee ↗ · NASA — the atmosphere & 100 km reference ↗
LIVE TELEMETRY / A CLOSED ENERGY BALANCE
Power in. Heat out.
T+ 00:00
Awaiting deployment. Launch the mission or go straight to orbit.
SOLAR ELECTRICITY
—Incident sunlight → DC busHEAT TO SPACE
—Radiator emission · infraredHOTTEST COMPONENT
—Throttle 70°C · trip 90°C · restart 68°CUSEFUL COMPUTE
—Fraction of nominal machine throughput01 / MISSION DESIGN
Change the balance.
Scenarios restart orbital telemetry. The shutdown scenario runs at full demand until its 90°C safety trip.
0° crosses the shadow center; high β can avoid it. This holds the Sun angle fixed, without seasonal drift or orbital precession. Changing altitude or β restarts telemetry.
No battery is modeled. In eclipse the compute core powers down, while stored heat continues radiating.
02 / THERMAL LOOP
Even space has a bottleneck.
Compute processors12.8 kW dynamic / 280 W·K⁻¹ to loop
Memory & interconnect3.2 kW dynamic / 160 W·K⁻¹ to loop
Coolant & radiatorFinite thermal mass / emissivity 0.90
Solar arrayInstantaneous equilibrium / panel inertia omitted
TEMPERATURE HISTORYLAST 30 SIMULATED MINUTES
● Processors● Memory● Radiator
Electronics heat—Environment absorbed—Heat stored / released—
Processor and memory heat conducts into a shared radiator. Infrared emission grows with absolute temperature to the fourth power.
03 / THE COST OF HEAT
Less work. Same overhead.
—per equivalent full-throughput compute hour
- Without thermal limits, same solar supply
- —
- Premium caused by heat
- Completed equivalent compute hours
- 0.000 h
- Accumulated operating cost
- $0.00
- Mission average / compute hour
- —
One compute hour means the work of this fictional 16 kW machine running at 100% for an hour. Throughput scales linearly with load. The fixed hourly cost accrues during cooldown and eclipse too.
CONTROL EVENTS
- Waiting for deployment.
ORBIT TRADEOFF / SUNLIGHT IS ONLY HALF THE STORY
Same machine. Two orbits.
Compare a 500 km day/night orbit (β = 0°) with a dawn–dusk sunlit-season approximation (β = 75°). Each starts from the same temperatures and uses your current solar array, radiator, demand, array alignment, thermal control, and assumed fixed cost. Playback position and test illumination do not affect these fresh runs.
The dawn–dusk preset has 100% sunlight in this fixed-angle run. Real dawn–dusk orbits can still have seasonal eclipses; this is not a year-round forecast. ESA’s Proba-2 orbit ↗
Preparing the orbit comparison…
TOTAL COMPUTE CHANGE
—MEAN HOTTEST COMPONENT CHANGE
—Dawn–dusk minus day/nightCOST PER COMPUTE HOUR CHANGE
—● Day / night● Dawn–dusk
An equivalent compute hour is the work of this machine at 100% for one hour. Flat segments mean no compute was delivered.
● Day / night● Dawn–dusk
Eclipse cools the day/night machine. Sustained solar power keeps the dawn–dusk machine warm and can increase throttling.
| Measure | Day / night500 km · β 0° | Dawn–dusk500 km · β 75° |
|---|
Total output and cost include eclipse, throttling, and shutdown. Operating cost continues during idle time. All hardware and costs are illustrative. Changing hardware recalculates both runs; selecting an orbit preset changes the live 3D mission.
Inside the model Equations, assumptions & sources +
Energy is accounted for.
Psolar = 1,361 W/m² × array area × cos(angle) × 30% cell efficiency × 94% bus efficiency. In shadow it is zero. Unused solar capacity is curtailed; conversion waste stays on the separately radiating array.
Qspace = εσA(Tradiator⁴ − 3 K⁴), with ε = 0.90. The three coupled nodes integrate C·dT/dt = power in − power out at steps no larger than 0.25 s. Thermal capacities: processors 18, memory 24, radiator 220 kJ/K. Chip and memory conductances are 280 and 160 W/K.
A simplified orbital environment.
Adjustable circular orbit from 500 to 35,786 km with Earth’s cylindrical shadow. Solar beta is the angle from the orbit plane to the Sun (distinct from array alignment). The Sun direction and beta are fixed during each run; seasons, precession, lunar shadows, and penumbra are omitted. At the default 500 km and β = 0°, the period is roughly 95 minutes. High beta can eliminate eclipse for the modeled geometry; this does not imply year-round eclipse avoidance in a real orbit. Absorbed radiator heating is an assumed 12 W/m² for Earth IR/albedo at 500 km, scaled by inverse squared orbital radius plus 0.12 absorptivity × 1.5% solar exposure in sunlight. Radiator area is total emitting area. There is no convective cooling, battery, active refrigeration, or radiator minimum-temperature heater.
Adaptive control reduces load above 70°C. Either processor or memory reaching 90°C forces sleep until both fall below 68°C. Awake housekeeping uses 800 W; sleep uses up to 200 W. The array estimate uses absorptivity 0.90 and two-sided emissivity 0.85. It is an instantaneous radiative equilibrium estimate, omitting panel thermal inertia.
An experiment, with stated limits.
The rocket integrates a configurable two-stage flight, with engine flow limits, changing propellant mass, gravity, drag, and simplified guidance. Arrival is not guaranteed. Its fictional hardware and control laws are educational assumptions, not a validated launch-vehicle forecast or launch-cost calculation. Spacecraft and arrows are exaggerated for readability; thermal color shows temperature, not visible incandescence. The fixed hourly cost is your assumption, not a market price or a hardware forecast. Launch costs, communications, radiation damage, degradation, and reliability are omitted.
Physics references: NASA — spacecraft thermal control ↗ and NASA — solar irradiance & Earth’s energy budget ↗. Orbit geometry: NASA thermal engineering guidebook ↗. The platform has four compute pods and two memory pods, with visible cold plates and supply/return plumbing. All compute pods share the model’s processor temperature and both memory pods share its memory temperature. The colored plumbing depicts the heat-transfer path; fluid flow and pump performance are not separately simulated. Visual design reference: NASA — station truss, arrays and radiators ↗. All hardware parameters and costs are illustrative design choices. Earth imagery: NASA/Goddard Space Flight Center Scientific Visualization Studio — Blue Marble ↗; data courtesy of Reto Stockli (NASA/GSFC) and NASA’s Earth Observatory.