Fuel and the torch drive
A torch drive is a rocket that throws water out of its nozzle at a very high speed, so a ship can push at about one g for days instead of coasting on a slow orbit.
33 numbers, 6 with an outside source
- A0 numbers at grade A
- B4 numbers at grade B
- C0 numbers at grade C
- D2 numbers at grade D
- E27 numbers at grade E
Functionality
3 topics
Range and the destination tags
- A trip burns the whole way while the fuel its BURN PROFILE spends allows, and coasts on the rest when it does not.
- It always keeps enough fuel to land and to break off one fight.
- A richer THRUSTER GRADE throws the same water faster, so a trip burns longer and coasts less.
- NEEDS FUEL | too little fuel aboard, so the coast runs longer than the stores. Refuel.
- OUT OF RANGE: cargo too heavy | even on a full tank the load makes the coast outlast the stores, so vent cargo.
- OUT OF RANGE: tank too small | even empty and full, the coast outlasts the stores, so a richer grade or a faster hull is the answer.
- OUT OF RANGE means no SINGLE crossing reaches it. It never means the place cannot be reached.
Burn profile
- A burn profile is how much speed one trip buys, chosen before you leave.
- Every profile flies at the drive's rated thrust, so the fuel spent is what changes.
- ECONOMY | the cheapest trip: the fuel bill plus each day's wages, upkeep and stores.
- FASTEST | all the fuel above the reserve.
- ARRIVE BY | the cheapest trip in all, fuel and days, that makes the day you name.
- Fuel you do not spend stays in the tank for the next leg.
- A slower trip eats more food and water and rolls for more contacts, and it can outlast the stores.
- A hull rated above 1 g strains its crew, shown as CREW STRAIN in fatigue per day.
Fuel and stores
- Fuel is water that the drive throws out the back, bought by the tonne at the station's water price.
- Food and water are scaled by the days and by how many people are aboard.
- CHOOSE THE WORK lists every service with its bill and its wait, and you pick how much of each.
- REFUEL, REPAIR, RESUPPLY and SERVICE do them one at a time.
Values
33 numbers: 4 B, 2 D, 27 E
Each value is read from the game code. Sort by any column heading, or filter by grade or by a word.
| Quantity | Value | Grade | Source | Why this value |
|---|---|---|---|---|
| Exhaust speed with no thruster grade fitted | 2,500 km/s | EAuthored | Atomic Rockets, enginelist2, Nuclear Salt Water (NSWR) | Derived by the game from a goal, that a full tank flies the close Mars run at one g; the published salt-water figure is 66 kilometers per second at 20 percent enrichment. |
| Exhaust speed of the top rung (TH-6) | 11,000 km/s | BSecondary | Atomic Rockets, enginelist3, Epstein Drive | The game takes the exhaust speed of the fictional Epstein drive from The Expanse. Atomic Rockets lists it as 11,000,000 meters per second (11,000 km/s), the game value, conjectured by its author from a mass ratio of 4 and a speed gain of 5 percent of light speed. The drive is fiction, so the claim is a fictional one. |
| Every rung of the exhaust ladder, no grade then TH-1 to TH-6 | 2,500, 3,200.2368, 4,096.6064, 5,244.0442, 6,712.8734, 8,593.114, 11,000 km/s | EAuthored | Atomic Rockets, enginelist2, Nuclear Salt Water (NSWR) | The ends are a derived base and the page's Epstein figure; the geometric spacing between them is the game's choice, and the ladder runs far past any published design on purpose. |
| Number of enrichment grades | 6 grades | EAuthored | No source recorded | The game chose six grades, TH-1 to TH-6; the setting canon says the recipe was frozen there. No source sets a number of grades. |
| Water in a full tank, as a share of the hull's dry mass | 1 share | EAuthored | Atomic Rockets, torchships, Torchship Performance (Rocketpunk Manifesto, On Torchships) | Authored by coldweld on 2026-09-28 as one fraction to measure later; the page's example torch designs use a mass ratio of 2 and its range runs up to 10. |
| Salt charge for one tonne of water at the base grade (the pad scales it with exhaust speed squared) | 10 Cr per tonne | EAuthored | No source recorded | A custody bill chosen by the game as the largest tested charge that stopped no starter, on a sweep too small to order nearby values; a physical price is unusable because the upper rungs need more energy than fission gives. |
| Burn a trip must always keep in reserve to break off one fight | 0.25 days of full thrust | EAuthored | No source recorded | Chosen by coldweld on 2026-09-28 as the burn that breaks off one fight, and kept when the water law arrived; no source gives a time to break off a fight. |
| Change of speed for a hop between two berths at one body | 4,040 m/s | DAnalogy | Atomic Rockets, appmissiontable, Terra-Luna space (the delta-v table, LEO to low lunar orbit) | The mirror's delta-v table gives 4.04 kilometers per second from low Earth orbit to low lunar orbit; the game uses that as the far end of what a hop between two berths at one body asks. |
| Change of speed to hold station at a body for a year | 100 m/s per year | DAnalogy | Atomic Rockets, appmissiontable, Stationkeeping | The table gives 30 to 100 meters per second a year for station-keeping at the L1 and L2 points and 50 to 55 for geostationary orbit; the game applies the top of that range to holding station by any body. |
| Acceleration a crew bears with no strain | 9.8 m/s^2 | BSecondary | Atomic Rockets, humanfactor, Table 1, gross effects of acceleration forces | The page's table of the gross effects of acceleration lists 1 g as Earth normal and 2 g as hands and feet heavy, with walking and climbing difficult. The game sets the no-strain line at 1 g (9.8 m/s^2), so a crew bears no strain below the first row that the table marks as an effect. |
| The three burn profiles | economy, fastest, arriveby profiles | EAuthored | No source recorded | Three burn profiles chosen by the game for play: the least running cost, all the spare water, or a named arrival day. The mirror ranks no burn profile. |
| Longest trip the game will quote | 36,525 days | EAuthored | No source recorded | A cap the game chose so a quote for an unreachable trip reads as no trip instead of dividing by zero; no source applies. |
| Highest grade a hull burns without custody rights taken (light craft) | 2 grade | EAuthored | No source recorded | Authored by the game and flagged for coldweld's review: the richest salt a hull outside the gram accounts may burn; the mirror has no rule for who may hold enrichment, because that is fiction. |
| Price of the TH-1 kit | 2,500 Cr | EAuthored | No source recorded | The game chose this price in credits; the price of every combat kit was multiplied by five together on 2026-09-21. No source prices a thruster kit. |
| Price of the TH-6 kit | 55,000 Cr | EAuthored | No source recorded | The game chose this price in credits; the price of every combat kit was multiplied by five together on 2026-09-21. No source prices a thruster kit. |
| Grade the TH-1 kit burns | 1 grade | EAuthored | No source recorded | Game structure: the first kit burns rung 1 of the exhaust ladder, and a grade sets the exhaust speed and adds no thrust. |
| Grade the TH-6 kit burns | 6 grade | EAuthored | No source recorded | Game structure: the top kit burns rung 6 of the exhaust ladder, and a grade sets the exhaust speed and adds no thrust. |
| Meters in one astronomical unit (beacon range and distance code) | 149,597,870,700 meters | BSecondary | Atomic Rockets, enginelist, the note that 1.0 astronomical units is defined as 149,597,870,700 meters | The mirror states that one astronomical unit is defined as 149,597,870,700 meters. |
| Meters in one astronomical unit (trip distance code) | 149,597,870,700 meters | BSecondary | Atomic Rockets, enginelist, the note that 1.0 astronomical units is defined as 149,597,870,700 meters | The mirror states that one astronomical unit is defined as 149,597,870,700 meters. |
| Fatigue a crew gains per day for each unit of acceleration above the no-strain line | 1.5 fatigue per day per m/s^2 | EAuthored | Atomic Rockets, humanfactor, the acceleration passage (the human body only can take 2-3 gees) | Chosen so that only the fastest hull nears crippling a crew on a long haul. The mirror says the body can take 2 to 3 gees for any length of time and gives no fatigue rate. The sources are silent on this figure. |
| Fatigue a crew recovers per day at or under the no-strain line | 5 fatigue per day | EAuthored | No source recorded | Chosen as the rest that balances the accrual rate. No source gives a recovery rate. |
| Points in one whole main-drive part | 100 points | EAuthored | No source recorded | The scale of a drive part at full repair. The points are the repair record, and a part performs by its state (see PART_STEPS). |
| Chance that a round through the stern reaches each main-drive part | nozzle, 0.5, chamber, 0.3, feed, 0.2 chance (nozzle, chamber, feed) | EAuthored | No source recorded | Authored in the game: the nozzle shows the most, so it is hit most. No source gives odds for a part of a drive. |
| Share of the hull's own health that a drive part takes before it is gone | 0.75 fraction of hull health | EAuthored | No source recorded | Authored so that a 44-damage round through a holed stern on a 180-point hull takes a third of a part. |
| Weight fleet doctrine gives a round that reaches the stern, for the main drive behind it | 0.5 weight | EAuthored | No source recorded | Authored: the stern holds the main drive, so doctrine counts a round there as costly (equipment.getpartexposure). A bow round counts 0. No source gives a weight. |
| Weight fleet doctrine gives a round that reaches a flank's jets | 0.1 weight | EAuthored | No source recorded | Authored: one jet cluster lost costs nothing, so doctrine barely counts it. |
| Share of thrust a wrecked nozzle still gives | 0.3 fraction | EAuthored | No source recorded | Authored so that a hull always limps home. No source gives the thrust of a damaged nozzle. |
| Performance of a drive part by state: fit, degraded, crippled, gone | 88, fit, 100, 41, degraded, 64, 1, crippled, 20, 0, gone, 0 percent by state | EAuthored | No source recorded | The states match the game's vital questions (fit over 87.5, degraded over 40, crippled over 0). A fit part works fully and the other states perform at their band's midpoint, so the average barely moves. |
| Most a crew with a welder can patch a drive part | 60 points | EAuthored | No source recorded | Authored: hands with a welder patch a part this far and no further, and a yard restores the rest. |
| Share of the fuel tank a wrecked feed line loses per travel day | 0.25 fraction of tank per day | EAuthored | Atomic Rockets, spacewardefense, the passage on a projectile hitting a tank of fluid | Authored. The mirror says a hit tank pumps its fluid out through the holes and gives no rate for a line. The sources are silent on this figure. |
| Share of the fuel tank a wrecked feed line loses per combat round | 0.02 fraction of tank per round | EAuthored | Atomic Rockets, spacewardefense, the passage on a projectile hitting a tank of fluid | Authored. The mirror gives no rate for a leaking line. The sources are silent on this figure. |
| Days a captain has to take up reactor custody rights from Safeguards | 90 days | EAuthored | No source recorded | A rule of the setting: Safeguards, the treaty regime, gives a captain this long to take the rights up. The shelf holds no source for it, so the figure is authored. |
| Hull classes that Safeguards does not account for by the gram | LIGHT CRAFT hull classes | EAuthored | No source recorded | A rule of the setting: only a light craft burns without a grant. The shelf holds no source for it. |
Real-world basis
8 points
- A rocket gains speed by throwing mass backward, so a faster exhaust gains more speed from the same mass.
- A torch drive has both high thrust and high exhaust speed, so it can push the whole way, flip at the midpoint and brake.
- A salt-water rocket dissolves a fissile salt in water, so the water is the propellant and the salt is the fuel.
- Zubrin's 20% enriched design gives an exhaust of 66 km/s, and a speculative 90% enriched design gives 4,725 km/s.
- Each U-235 fission releases a fixed 202.5 MeV, so a faster exhaust needs more fissions in every tonne.
- The energy in a tonne of exhaust grows with the square of its speed, which is why the salt charge scales that way.
- A ship that does not need to hurry can burn, drift for most of the distance and brake at the end.
- Thrust stays fixed while the tank empties, so acceleration rises as the ship gets lighter.
Game and reality differences
8 points
- The ladder runs far past a published design. The real salt-water figures are 66 km/s and, in the speculative case, 4,725 km/s. coldweld chose to keep one-g trips, so the ladder is a labeled fiction.
- The top rung is the Epstein figure, a fiction on the mirror's own page. Robinson calls an exhaust that fast wasteful for most solar system routes; the game keeps it so a torch can fly one-g trips.
- The base rung is derived so a full tank flies the close Mars run at one g. The spacing of the middle rungs is authored.
- A thruster grade changes only the exhaust speed and adds no thrust. The old design added acceleration and was removed.
- A grade above the base burns only while the hull's custody rights are taken. Safeguards holds every gram of fissile material, and nobody can grant it.
- The salt charge is a custody bill, scaled by exhaust speed squared. A physical price was unusable because the upper rungs need more energy than fission can give.
- Gravity and orbits are ignored in the trip time; a trip is a brachistochrone, or burn, coast and brake when the water is short.
- The tank holds water equal to an authored 100 percent of the dry mass, so with an empty hold the mass ratio is 1 plus that share. The mirror's torch designs run from 2 to 10, so the game takes the low end.
Source absences
6 points
- No source on the shelf provides values for the following:
- Price of salt, custody, or a refill. Every Cr value is authored.
- Tank water against dry mass. The tank share is authored.
- Reserve to land.
- Reserve to break off a fight. 0.25 days of full thrust is authored.
- Atomic Rockets does not cover the following:
- Who may hold each enrichment grade. The rule is fiction.
- Rank of burn profiles, cost-to-arrive plans, or arrive-by days.
Influences
3 works
- Nuclear salt-water rocket (Robert Zubrin, 1991) (real system): the fuel itself: fissile salt dissolved in water, with the water as the propellant
- The Expanse (Epstein drive) (book): the top exhaust speed of the thruster ladder, taken from the mirror listing of the fictional drive
- Sergei Korolev (person): the title Chief Designer, given to the unnamed designer of the torch drive in the setting canon
Sources
7 documents
- Atomic Rockets, enginelist2, Nuclear Salt Water (NSWR): A 2% solution of uranium tetrabromide in water. At 20% enrichment the exhaust speed is 66,000 m/s (6,728 s). The 90% enriched case lists 4,725,000 m/s. The page says the water is the remass. "considered far-fetched by many scientists" (all uses)
- Atomic Rockets, torchships, Introduction: A torch drive has high thrust and high specific impulse, so it can fly a brachistochrone. Heinlein coined the word in 1953 for his story Sky Lift. (all uses)
- Atomic Rockets, torchships, Torchship Performance (Rocketpunk Manifesto, On Torchships): A 1,000 ton example with a mass ratio of 2 and a 300 km/s exhaust gets about 200 km/s of speed change. It shows a tank equal to the dry mass is a plausible design. (all uses)
- Atomic Rockets, torchships, Constant-acceleration Delta-V: At a fixed thrust the acceleration rises as the mass falls. The page says to cut thrust to hold the acceleration steady. (all uses)
- Atomic Rockets, torchships, Low Gear / High Gear (Robinson and Campbell): A 1,000 ton ship with a 10,000 km/s exhaust crosses 1 AU in 10.5 days, flipping at the midpoint, and burns 66 tons. A variant accelerates for a tenth of the distance, drifts for eight tenths and brakes for a tenth. Robinson says a 10,000 km/s exhaust is wasteful for most routes in the solar system. (all uses)
- Atomic Rockets, enginelist3, Epstein Drive: A fictional fusion drive from The Expanse, listed at 11,000,000 m/s exhaust speed and 1,100,000 s of specific impulse. The 11,000 km/s figure is the page's conjecture from a mass ratio of 4. (all uses)
- Atomic Rockets, atomicfuel, The Details: One U-235 fission gives 202.5 MeV, which is 3.244e-11 joules per atom. (all uses)