Last time, we talked about FTL travel. However, to be able to FTL-travel, we still need to leave that pesky gravity well behind - and at roughly 9.4km/s, it’s a doozy. Just looking at today’s rockets, mass fractions and propellant densities make multi-stage rockets necessary. But how does it look like in the future?

Shuttles and Engine Technology

From First Principles: Engine Technology

Looking at GURPS Spaceships, p. 21ff, we do have a few reaction engines available. Let’s look at those with sufficiently high Thrust-to-Weight Ratio (TWR) in more detail:

Type Fuel g per system delta V per tank (km/s) Isp (s)
Chemical Hydrolox 3 0.225 440
Chemical Kerolox 6 0.18 350
HEDM Metallic H2 2 0.75 1500
NTR Hydrogen 0.5 0.675 1300
NTR Water 1.5 0.225 440
Orion Drive Nuclear Bombs 2 12 24000
Fusion Torch (^) Hydrogen 0.5 22.5 44000
Fusion Torch (^, HT) Hydrogen 1 11.25 22000
Fusion Torch (^) Water 1.5 7.5 15000
Fusion Torch (^, HT) Water 3 3.75 7500


The hydrogen-using version has been reverse-engineered directly from GURPS Spaceships; the numbers are suspiciously identical to LH/LOX rockets. I’ve therefore created a kerosene-using rocket which mirrors the real-life rockets. The theoretical ISP of 350 implies 0.18km/s for each fuel tank of rocket fuel (probably RP-1 and LOX). It also features a significantly higher TWR than Hydrolox, again mirroring real-life engines.

The disadvantage of both is the rather horrible efficiency - to take off from Earth, a hydrolox rocket has to spend almost eighteen fuel tanks to take off, leaving a total of a) one system for armour, b) one-third system for the (smaller-sized) control room, c) one-third system for the engine (1G; it only begins to move once burning some fuel) and d) one-third system of cargo - meaning the payload fraction is a whooping 1.6%. A kerolox rocket is even worse - it has space for one and a third systems, of which one is reserved for armour, one-third for the engines (2G), and we just ran out of space for control room or payload - that one requires multi-stage rockets.

Summary: Hydrolox requires spending about 55 tons of hydrolox per ton of cargo to orbit, for an operating cost of roughly $44k per ton to orbit. This is significantly cheaper than today; but that’s because we’re not counting the cost of the rocket - roughly $770k for a 100t rocket, with a payload of 1.6t, for a total cost of $525k per ton to orbit - significantly cheaper than today, but that assumes mass production of rockets.


HEDM is the newer, cooler and more powerful rocket for the future. GURPS Spaceships talks about unspecified high-energy density materials, and lists two examples: Metallic hydrogen and metastable helium. The ISP of the default HEDM pegs that one as Metallic Hydrogen; Atomic Rockets puts it at pretty much the identical value. Unfortunately, metastable helium has the disadvantage of not reaching a TWR bigger than two (i.e. 0.1G per system) and the slight disadvantage of probably spontaneously exploding after about two hours. Therefore, Metallic Hydrogen.

An HEDM-fuelled rocket has ten and a third systems left-over after fuel. One system for armour, two-thirds of a system for the engine (which gives 1.33g acceleration). If we assign a full control room, that leaves us with eight and two-third systems of payload. For an SM+6 100t shuttle, that means having spent about 49t of fuel, and get almost 44t (or 52 passengers) of payload into orbit. Total cost in fuel is $6.7k per ton of cargo.

Summary: HEDM engines cost about $6.7k per ton to orbit in fuel alone, spending slightly more than one ton of fuel per ton of cargo.

Nuclear Thermal Rocket

An NTR takes fuel, heats it in a nuclear reactor, then expels it. This exhaust has the slight issue of being radioactive. It also has less ISP and TWR than an HEDM engine. Its one advantage is that the fuel is really cheap - one third of the cost of HEDM per ton.

An NTR with hydrogen has nine and two-thirds of a system left over after fuel. One for armour, one for the control room and three for the engines (1.5g) leaves four and two-thirds for payload. That’s 2.2 tons fuel per ton payload, or $4.4k per ton. Not that much cheaper than HEDM, considering its exhaust is radioactive.

However, there is an alternative: NTRs can use the Ram-Rocket option (see GURPS Spaceships, p. 30), which means they do not require fuel while in atmosphere. This means we can subtract the maximum speed in atmosphere (4,500km/h for a hydrogen-fuelled NTR; that’s 1.25km/s) from the actually required delta v, for only 8.15km/s needed. This saves another system of fuel; fuel is now 1.65 tons per payload-ton, for $3.3k per ton.

Summary: NTRs are less powerful and less efficient than HEDM rockets. In the end, they have a slightly lower cost at $3.3k per ton and aren’t volatile, but pay with their radioactive exhaust.

Orion Drive

What could conceivably be called the space nerds’ wet dream: Propulsion through superior firepower.

A hypothetical TL10 orion spacecraft, at 10,000t, would mount three nanocomposite armour systems (one on the front, two as a pusher plate), one control room and one engine system (2g). It would need 400t of nuclear propulsion units per launch, less than one system. That leaves a staggering fourteen systems for cargo; that’s 7,000 tons. On the other hand, the nuclear bomb units are expensive at $250k per ton. Meaning our single launch costs a hundred million dollars, or $14k per ton to orbit.

All of this is discounting the many problems of an orion drive. Possibly destroying a starting pad, nuclear fallout, and EMP effects in the upper atmosphere.

Summary: With the main advantage of orion drives (being their early availability) negated, they do not make economical sense in the setting, especially considering their disadvantages.

Fusion Torch

The Fusion Torch is the only superscience engine technology on my list. That’s because it is limited superscience, i.e. restricted by engineering issues, not by fundamental science. And it truly is awesome: 22.5km/s delta v per tank, with sufficient TWR to take off from a planet.

The first configuration we’re looking at is using Hydrogen. It needs three systems of engine, one control room, one armour system and less than half a system of fuel to take off. Let’s round this up to one system, leaving us with 14 systems for payload. For an SM+8 spacecraft, that’s 25 tons of hydrogen (for $50k) used for 700t of cargo, for a cost to orbit of slightly over $70 per ton - by far the cheapest option yet.

The other configuration is going for the even cheaper propellant. A water-based rocket gets more thrust, but less ISP. In this case, this translates to only needing one engine system, but 1.25 fuel systems per launch. Calling this 2 systems, we end up with 15 payload systems (750t), and using 62.5t of fuel per launch. On the other hand, that propellant is cheap as (literally) water, clocking in at $1250 per launch, or basically nothing (less than $2) per ton. At that point, any other cost is going to be far more important - like the capital investment costs for the roughly $32.6M shuttle, which is about $9k assuming one launch per day and ten years of use, maintenance, personnel costs, and freight handling costs. This pushes us to about $20 per ton cost, and probably about $30 per ton for the end user.

However, it does have a disadvantage. The Designer Notes, which clarifies some things on the exhaust of drives, tells us that there’s slightly radioactive exhaust.

By the Books: Spaceships 2

Now that we have arrived at costs from the bottom-up, so to speak, we can take a look at canonical GURPS prices. GURPS Spaceships 2 p. 40 gives us interface rates per ton of cargo. How do we compare to them?

TL 7/8 Reaction Drives

$300k per ton as given, compared to our values of $44k per ton for fuel, or $525k per ton when throwing away rockets. That’s with single-stage rockets. If we look at the Baikonur Launch Vehicle from GURPS Spaceships 2, p. 17, it lifts a thirty-ton payload in hangar bays (which would be fifty tons of pure cargo) for $10.7M. That’s $214k per ton (throwing away everything). All of those are close enough that we can assume $300k per ton to be a good value for those drives.

TL 9 Reaction Drives (HEDM)

Those are HEDM rockets. $50k per ton is given in Spaceships 2; our own rates are $6700 per ton in fuel costs. That’s a larger difference, and one that can’t be explained entirely by loading and unloading or infrastructure expenses. Looking at the Conestoga HLV (GURPS Spaceships 2, p. 18), it launches two systems of cargo (100t if using only cargo) for $3.6M in fuel costs. If we assume the rocket to be reusable, that’s $36,000 per ton in fuel costs, which is much closer to the $50k above. What causes this cost discrepancy? First of all, the Conestoga has three systems of armour instead of one. That’s two more cargo systems. Additionally, they use two engines instead of two thirds of a system. That’s another one and a third system. Additionally, they use twelve instead of nine and two-third systems for fuel. That’s another two and a third system, for a total of five and two-thirds additional cargo systems, or almost four times the capacity!

Torch Drives

They can be summarized as “cheap as hell” - $50 per ton to get into orbit. My own numbers give me $70 per ton for hydrogen-fuelled, and less than $2 per ton for water-fuelled. Apparently, Spaceship’s numbers are for a water-propelled spacecraft. The cost difference is explained by docking fees, infrastructure etc. Also by adding some safety measures, whose mass penalty one can easily accept.


What about the alternative? A space elevator, while impossible with today’s technology, seems achievable with TL9 or TL10 technology. Ultratech (p. 224 pegs it as $3 per lb, or $6,000 per ton - that’s almost as much as using HEDM rockets, and more expensive than an NTR. Assuming limited superscience, the peanuts-per-ton water fusion drive blows it away. And that’s discounting capital investment costs, which’d add another million dollars per ton (!) if paid off over a hundred years (!).

And that’s where one of the issues with UT comes in: We don’t have a clue on how those numbers were arrived at. According to these rules, it costs the same whether built at TL9 or TL12. It costs the same whether you have or don’t have superscience. That’s annoying.

So we’re looking at some other sources. A presentation by Bradley Edwards from 2003 gives the cost for their elevator at roughly $7B for a capacity of 5t per day - that’s less than four percent of UT’s cost, or $76k per ton when paid off over 50 years. However, most of that’s made up of launch costs ($1B), a tracking station ($500M), and power beaming stations ($1.5B). Their prediction is for the second cable to cost $2B, or $22k per ton over fifty years. Still extremely expensive, though less so than before. It’s actually competitive with chemical rockets.

How can we reduce that further? Looking at their numbers, we can assume that many things are reduced by TL: The launch costs should be reduced to almost-zero through bootstrapping or torch drives. Cable, spacecraft and climber production should scale similarly. We’re going to need a tracking facility anyway. Power beaming stations should either be cheaper or may even be eliminated by on-board storage. I’m also going to ignore the contingency, since it’s going to amortize over many elevators. Assuming the cost is reduced by 75% at TL10, that leaves us with a cost of $160M per ton per day, or $9k per ton over fifty years. Now it’s actually competitive with HEDM rockets - but not by much.

On the other hand, Spaceships 2, p. 40 gives us $10,000 for a TL9 space elevator, $100 for a TL10 space elevator, and $10 for a TL10^ one. If we assume this to include amortizing, our numbers above can be interpreted as pertaining to a TL9 elevator - either with a longer amortizement period, or with the reduced-cost I had originally assumed to occur at TL10.

On the other hand, the $100 implies a reduction in cost by a factor of 100. Whether that’s due to having paid off the construction costs or improvements in material science and space technology is something to be decided. Personally, I kind of like the thought of space elevators as a reminder of a time where no torch drive was available - only remaining economically competitive with large, long-term throughput.

Laser Launch

Laser Launch, or That Thing Transhuman Space Does, is heating some sort of ablative plastic using ground-based lasers. The advantage is that it’s cheap as hell - the plastic costs $80 per ton while providing the same dV as a tank full of HEDM fuel for $6k per ton. However, you do need ground-based laser stations to accelerate the craft - 30GJ of laser stations for SM+5, increasing linearly with mass. That is going to be expensive. Assuming we build a small-ish laser station capable of lifting an SM+6 craft, it’s going to cost roughly $3B, and mass 30,000 tons.

But how expensive is a launch using those? If we take the Mercury HLV from Spaceships 8, p. 17 at SM+9, we get a cargo capacity of 1400t of cargo. Assuming that one can simply combine laser stations, one would need 30 of the above to lift the Mercury, for an investment cost of $90B. Assuming one launch per day, that’s $3.5k per ton amortized over fifty years. Of course, we can launch much, much more - a single launch takes about five minutes to Earth orbit. We could launch almost 150 Mercuries per day. Assuming a more conservative ten launches a day, we’re going to have to pay $350 per ton to amortize, plus $150 per ton for fuel costs - it also launches 14,000 tons per day, or more than five million tons per year out of Earth orbit. That corresponds to 500,000 people.

How does that compare to the official numbers? Transhuman Space: Changing Times, p. 23 gives cost to orbit for a passenger with $24,500. Since passengers cost 10x the freight cost, that gives us a per-ton cost of $2,450. That’s slightly more expensive than what I had assumed above, but is probably due to the smaller scale.


Looking at the available methods, it is clear that there’s one decision left to take: Whether to allow the fusion torch engine or not.

If it is not available, the cheapest launch method is the laser launch, at about $1,000 per ton. The most flexible is an HEDM engine at about $7,500 per ton. A space elevator would be competitive if we’re comparing book prices.

If, however, there are fusion torches, there’s no contest. Fusion torches (water-propelled) are cheaper than everything else. However, there’s still to be decided whether to include them in the setting - at 22.5km/s delta v per tank, you’re not really restricted by delta v anymore. I’ll check in the next post: Space propulsion.