The data in words
A vertical cross-section, altitudes to scale up to 80 km. A Thermal Chimney rises from 1 km to 65 km. Fans at its base pull 737 K air through heat exchangers at 20 m/s, cooling it by 300 K; the cooled air drains into a lowland basin. Sealed liquid loops, stage by stage, lift 1.2 × 10¹⁵ W per chimney to the Exhaust canopy, 80 km across, whose tiles fire 2.3 µm light into 25° cones, 3.0 × 10¹⁵ W per canopy. The clouds lie between 48 km and 70 km. The Braking Ring at 400 km takes the Lattice's beam at 10 MW/m², sheds its receiver loss heat at 1,500 K, and sends power down tethers to the canopies. In the basin, after 2036, dry ice 2 km thick at 180 K lies under a cap of 2 m of basalt, 20 m of aerogel and a 1 mm membrane; heat pipes 200 m into the floor carry 12 PW at first up a riser to radiators on the cap.
| Element | Where | What it does |
|---|---|---|
| Thermal Chimney | 1 km to 65 km | lifts 1.2 × 10¹⁵ W in sealed liquid loops |
| Intake fans and exchangers | from 1 km | cool 737 K air by 300 K |
| Exhaust canopy | 65 km | throws 3.0 × 10¹⁵ W to space at 2.3 µm |
| Clouds | 48 km to 70 km | sulfuric acid |
| Braking Ring | 400 km | takes the Lattice beam at 10 MW/m²; power down tethers |
| Basal heat pipes | 200 m into the basin floor | carry 12 PW at first |
Principle
A radiator is not enough. Anything radiates at most σT⁴ at its own temperature. Venus, shaded, radiates from its cloud tops at about , in all, a few hundred watts per square meter. At that rate cooling would take . To finish in a few years, heat has to leave far faster than any surface at the air's temperature can shine. It has to be pumped.
A light-emitting heat pump. An LED makes about one photon per electron. The photon's energy is set by the semiconductor's bandgap; the electron's energy is set by the voltage. Drive the diode just below its bandgap and each photon leaves with more energy than its electron brought. The difference is taken from the vibrations of the crystal, which is to say from heat: the LED cools while it shines. MIT showed it, an LED emitting of light from of electricity (Santhanam 2012). The Exhaust's tiles must do the same at a power density that has never been approached, with a bandgap of , which sets the wavelength.
LED light is not heat radiation, so it is not limited by σT⁴. The voltage packs photons into the diode's band far more densely than a thermal body at the tile's temperature would. That lets a tile at a few hundred kelvin shine , what a black body at radiates.
The Carnot cost. Nothing escapes the second law. A heat pump moving heat dQ from a cold side at T_c into a hot
reservoir at T_h needs at least dW = dQ (T_h / T_c − 1) of work. Here the cold side is the air, and the hot
reservoir is the light itself; the model takes its temperature as the emitters' effective
. Summed along the cooling path, from
down to :
- the sensible cooling costs of work;
- the latent heat, released as the CO₂ condenses and freezes between and , costs , more than the sensible part because it comes out when the air is cold and every joule is dearest;
- in all, for of heat: joules of electricity per joule moved.
Heat and work both leave as light, in all. The model checks that the pumps never beat Carnot (check). It charges exactly the Carnot work, the least the second law allows; real tiles will need more.
The trade. The effective temperature is a choice. Hotter light means brighter tiles and less area, but more work per joule. Run the same campaign with the coolest emitters the model tries and the working emitters would cover of Venus; with the middle option, ; with the hottest, , for more work. The ledger gives each option's temperature. The plan picks .
Against a shaded Venus radiating alone, the Exhaust throws out times as much light on average, and removes times as much of the planet's own heat. The difference between the two is the work.
Design
Canopies. Each canopy is a hexagonal aerostat across, , floating at the top of its Thermal Chimney, up. It floats on CO₂ warmed by the heat arriving from below, a hot-air balloon heated by the heat it is pumping out. Exchangers in its floor take heat from the chimney's liquid loops; tiles on its top throw it to space. As the CO₂ rains out and the air thins, the canopies float lower. On average each one takes in of electricity and throws out of light.
Tiles. Each tile is an InGaAsSb thermophotonic diode with a bandgap. A photon-recycling cavity sends light that would be reabsorbed back for another chance to escape. Every electron that recombines without making a photon dumps its energy into the tile as heat, so the tiles must reach an external quantum efficiency of . A photonic-crystal face aims the light upward into a cone.
Why . The thin CO₂ above the canopies barely absorbs at that wavelength, and the Hesperus Veil is made transparent there, so the beams leave Venus and pass the Veil untouched.
Sizing, and why version 3 made it bigger. Version 2 sized the Exhaust from the campaign's average, , which of working tiles can shine. The day-by-day simulation shows that the average hides three different phases:
- January to June 2033. The Lattice is still growing and all of it goes to cooling. Cooling is fast.
- June 2033 to August 2034. The Ceres Railheads draw of the Lattice's . Cooling crawls; the Clear Rain begins on .
- From August 2034. The Ring recovers from the arriving ice and powers the Railheads. Power is now plentiful, and the emitters, not the supply, set the pace.
So the emitters must be sized for the peak of the third phase. With version 2's smaller Exhaust the simulated air becomes Earth-like months after the deadline. With of emitters, , the CO₂ is down by and the air is Earth-like on , with to spare. A bigger Exhaust buys nothing: the finish then waits on the Forges' oxygen, and the extra pumping overdraws the Lattice.
The data in words
Temperature falls from 737 K to the CO₂ frost point and rises to 288 K; surface pressure falls to 2.4 bar, with 0.25 bar of oxygen and 310 ppm of CO₂.
| Event | Date |
|---|---|
| The Clear Rain begins | 19 January 2034 |
| The CO₂ seas freeze | 14 March 2036 |
| All the CO₂ is down | 20 April 2036 |
| Frost Towers finish | 9 May 2036 |
| Caps sealed, the Veil opens | 30 June 2036 |
| Earth-like | 30 September 2036 |
| Deadline | 31 December 2036 |
What the simulation pumps. The simulation spends of pump work, more than the the air alone needs. Heat keeps arriving from the crust, the Forges' waste, the falling ice and the sunlight the closed Veil lets through, in all, much of it late, when pumping is dearest. With everything else, the plan's demand is against the Lattice's : to spare from the Lattice alone, once the Ring powers the Railheads.
Frost Towers. At the air still holds of CO₂ vapor, . The same tiles on surface towers held at freeze it out down to , leaving airborne and in the finished air. That takes of work and is done by .
The data in words
A CO₂ phase diagram, pressure on a log scale against temperature, with the sublimation, vapour and melting curves. The simulated path of the air's CO₂ runs from Venus today (737 K, 90 bar of CO₂) to Earth-like on 30 September 2036 (288 K, 74 Pa of CO₂). The sealed dry ice sits at 180 K, from 2.4 bar at its top to 280 bar at its base, where it would melt at 222 K.
| Point | When | State |
|---|---|---|
| Venus today | Before cooling | 737 K, 90 bar of CO₂ |
| Critical point: the Clear Rain begins | 19 January 2034 | 304 K, 73.8 bar |
| Triple point: the seas freeze | 14 March 2036 | 216.6 K, 5.2 bar |
| All the CO₂ is down | 20 April 2036 | 195 K, 1 bar |
| Frost Towers finish | 9 May 2036 | 74 Pa of CO₂ left in the air |
| Frost Tower cold trap | 133 K, 74 Pa | |
| Earth-like | 30 September 2036 | 288 K, 74 Pa of CO₂ |
| Sealed dry ice | Sealed by 30 June 2036 | 180 K, 280 bar at the base; melts there at 222 K |
Specifications
| Parameter | Value |
|---|---|
| Canopies | , each about across |
| Emitter area | working at the average, installed for the peak |
| Emission | at , in a cone |
| Effective emitter temperature | |
| Heat removed from Venus's air | (average ) |
| Pump work for that heat | , J per joule moved |
| Pump work in the simulation | , counting heat from the crust, the Forges and the ice |
| Peak power out | |
| Per canopy | in, out |
| Frost Towers | Tiles at trap the last CO₂ to |
What is invented
Build sequence
- : Exhaust-1 begins in the laboratory. Its kill signal, no net cooling at useful power, would send the plan back to the multi-century route.
- : Exhaust-1 must have passed.
- : Under the closed Veil, the canopies float up and their chimneys are lowered.
- : Active cooling begins, on the growing Lattice's power.
- : The Railheads take most of the power; cooling slows.
- : The Clear Rain: liquid CO₂ starts to fall.
- : The Ring pays for the Railheads. The Exhaust runs at its limit from here on.
- : The CO₂ seas start to freeze.
- : All the CO₂ is down. The Frost Towers take over.
- : The CO₂ left in the air is at its target.
- : Active cooling ends; the caps are sealed and the Veil opens.
- : Venus is Earth-like.
Failure modes and safeguards
- Dead tiles. A failed tile goes dark and its neighbors take its share. Installed area, , is well above the working on average.
- The Ring overhead. The Braking Ring circles up. Tiles whose cones would cross it stay dark.
- Stray light. Each canopy's light spreads through its cone. Long before it reaches the Veil it is far fainter than sunlight, and the Veil is transparent to it anyway.
- A stalled Exhaust. Under the closed Veil, Venus gains heat only slowly, from the crust and the little sunlight the Veil lets through, so a pause costs time, not ground.
- The second law. The model charges ideal pumping. Every real loss, and the entropy of narrow, beamed light, adds work, and from the Lattice alone the budget has only to spare. The Ring's recovered energy is the real margin.
Reproduce these numbers
The Exhaust's numbers come from calc/src/model/exhaust.ts; the day-by-day cooling from calc/src/model/trajectory.ts.
pnpm calc --only exhaust-work-ratio # Carnot work along the cooling path, per joule moved
pnpm calc --only exhaust-area-installed # peak power / σT⁴
pnpm calc --only traj-earth-like # the day-by-day simulation's finish
pnpm calc --only traj-pump-energy # what the simulation actually spends
In the Explorer, lower exhaust-peak toward version 2's size and the simulated
finish slips past the deadline. Raise it and the finish does not move, because the oxygen sets it, while the pump work
and the energy demand rise until the Lattice alone no longer covers them. Lower exhaust-effective-temp and each joule
costs less work but the emitter area grows steeply; raise it and the area shrinks but the work grows.