Assuming it would be possible (but not required) to share resources between ships during the journey, but doing so takes minimal (but non-zero) resources.

Is it more fuel/space efficient to have a fleet of generation ships (either several self-contained ships, or specialised ships), or to have one large ship for all functions (self-contained)?

  • 43
    $\begingroup$ Redundancy has value. It is cheaper to buy one large hard disk than several smaller ones plus a RAID controller; but storing all your precious data on one large hard disk is risky, because if that hard disk fails you lose it all. Sending all your colonists on one large ship is more risky than dividing them among several smaller ships. The decision of how much redundancy is needed or recommended cannot be made without a set of defined values and a risk assessment: which the question conspicuosly lacks. $\endgroup$
    – AlexP
    Dec 1, 2020 at 12:46
  • 8
    $\begingroup$ @alexp that's because I'm not asking "How much redundancy do I need in a fleet of generation ships?" (That's a good next question though). Your comment-answer of "Sending all your colonists on one large ship is more risky than dividing them among several smaller ships." Is a good start, and I encourage you to post it as a fully fledged answer. $\endgroup$ Dec 1, 2020 at 12:51
  • 2
    $\begingroup$ *Sending all your colonists on one large ship is more risky than dividing them among several smaller ships." Is a good start, and I encourage you to post it as a fully fledged answer. — but that's not what you asked in your original question. You are changing the question, which isn't a good thing to do on SE sites. ¶ @Ash has answered the original question. The other two answers are good, but they answer a different question. $\endgroup$ Dec 1, 2020 at 15:17
  • 3
    $\begingroup$ Why choose? Make the ships modular, then bolt them into a giant hive/bigger ships as appropriate. If you make them giant cubes, perhaps. Give all your citizens cybernetic uplinks to a computer that can direct repairs and module reassembly. Forcibly steal useful technologies from other races. Assimilate their crew. Add their biological and technological distinctiveness to your own.... wait.. I think I got sidetracked... $\endgroup$
    – Joe Bloggs
    Dec 2, 2020 at 16:45
  • 2
    $\begingroup$ @AlexP: Just a minor quibble here. It's not necessarily cheaper to buy one large hard disk rather than several smaller ones, since manufacturing the smaller ones brings in economies of scale. Indeed, the RAID acronym comes from "Redundant Array of INEXPENSIVE Disks". We might likewise expect economies of scale in building many smaller generation ships rather than one large one. $\endgroup$
    – jamesqf
    Dec 2, 2020 at 17:17

16 Answers 16


It's more efficient to send one big one

Running 10 small sewage processing plants uses more power than 1 big one. Same with maintaining 1 big nuclear reactor vs maintaining 10 small ones. Same with Co2 filters, Gyms, Schools, Creche, etc. Are you going to build 10 playgrounds for children? Do you have 10 autopilots, as well as 10 navigation computers, 10 gyroscopes, 10 airlocks? As well as mandatory redundancies for all these systems. Do you have 10 kitchens and 10 dining rooms?

10 small ships also burn through spare parts faster than 1 larger one. Spare parts (storage or manufacturing) will be one of your biggest logistical challenges. 10 ships with 10 x 2-core nuclear reactors (you need a spare when one is being maintained) will burn through parts faster than 1 x 4-core reactor.

For 8 hour shifts, 40 hour weeks, with sick leave and 2 weeks holiday a year, a position that needs 24/7 coverage will need 5 fold staffing to ensure that a post is always manned. So you'll need 5 x 10 = 50 people to ensure that each ship has someone on duty. So 50 people capable of piloting the a ship. 50 trained nuclear techs. 50 police officers (minimum). 1 big ship may need 2 or 3 concurrent police officers on duty at any one time - but you'll only need a staff of 10-15 total officers on the roster in order to get 2-3 on duty at any time. That simplifies your training regime considerably.

If you have shipbuilding facilities that can only build a ship up to a maximum size it may be tempting to build small generation ships for a fleet - to which I'd suggest build the ship in sections and join it together in a vacuum.

Is there a middle ground between one big ship catastrophically failing and lots of little ships with totally redundant systems?

I note in the comments (now moved to chat) there's a discussion about "It's not very efficient if your only ship blows up". So as an aside this can be settled too:

Redundancy vs efficiency is a tricky problem. However there is a middle ground for this problem.

  • For everything you're bringing on your journey, every compartment type, facility, station, or piece of equipment, give it a random number between 0 and 1. Eg main engines are 0.423421, hydroponic tomato growing is 0.1267542, storage of toiletries is 0.854223, sewage processing is 0.73323, primary school classrooms are 0.5422, accommodation for families (named A...F) is 0.52321, accommodation for singles (named S...Z) is 0.2214, nuclear reactor is 0.14321, etc. This list should have thousands of entries.
  • Build N ships, giving each a number K from 0 to N. Ship K has all the equipment between ((K-2)/N modulo 1) to ((K+2)/N modulo 1). So ship 4 of 10 has everything between 0.2 and 0.6. Ship 0 of 10 (the first) has everything between 0.8-1.0 and 0.0-0.2.
  • Get your ships into orbit, rig them together in a strong frame (in a random order - not sequential), build conduits for power/air/water/etc between them that can be disconnected in an emergency, and build hallways between them that can be sealed in an emergency.
  • Now you you have one big ship made up of a fleet of smaller ships, where every system is replicated 4 times and no more.
  • If one ship has a catastrophic failure and blows up, every system that was lost has 3 other redundancies.
  • Now you can lose any 3 ships from your 10 without compromising the mission.
  • Losing 4-7 ships may compromise the mission, but it may be possible to finish the mission with every service available with only 3 ships.
  • You can tweak N if you want more or less ships. Change the +2 to a bigger or smaller number if you want more or less redundancy.

(For those of you who spend your nights clicking through random Wikipedia pages - this is a DHT)

  • $\begingroup$ Comments are not for extended discussion; this conversation has been moved to chat. $\endgroup$
    – L.Dutch
    Dec 3, 2020 at 17:03
  • $\begingroup$ You can als add a bit about Shielding/Weight/Materials. A single big Sphere has a lot less surface-area and will thus have a vastly better ratio of material&weight for the same amount of carrying capacity. - And you will need quite good isolation to keep the heat in, radiation out and protect from interstellar debris. A single big ship is vastly more efficient in this regard, too. $\endgroup$
    – Falco
    Dec 4, 2020 at 8:20
  • $\begingroup$ 0 to N gives you N+1 ships $\endgroup$
    – Andreas
    Dec 5, 2020 at 6:34

As already pointed out in other answers/comment, a single big ship is more efficient BUT will increase the chances that a single, catastrophic failure leads the entire mission to failure.

A reasonable compromise seems then to use a few large ships instead of a single, humongous one. The individual ship will still be big enough to cash on the scale efficiencies, and the redundancy implicit in having two or more of them increase the chances that at least one of them will make it to the destination.

More or less what NASA did when sending out the planetary probes in the 70's, with multiple Pioneer, Voyager and Mariner probes being sent out in very close intervals.

  • 3
    $\begingroup$ Thanks for tying it to actual NASA missions, that's really helpful. $\endgroup$ Dec 1, 2020 at 13:11
  • 1
    $\begingroup$ Advantage of having large ships is also having more recreational space, or at the very least more area to move to/through. This can help very much with the psychological stress of being locked on a relatively small area for longer periods. $\endgroup$
    – Trioxidane
    Dec 1, 2020 at 14:38
  • 4
    $\begingroup$ This is what the early explorers did. They took small fleets of 3+ vessels in search of new lands. en.wikipedia.org/wiki/… They had lots of ship failures, and were mostly quite glad that they had more than one ship. $\endgroup$
    – UEFI
    Dec 2, 2020 at 10:31
  • 3
    $\begingroup$ "a single big ship is more efficient" is a bad generalization and not always true. There are a lot of limiting factors, especially engineering challenges, when anything gets bigger. $\endgroup$
    – LukeN
    Dec 2, 2020 at 17:55

If shielding from cosmic radiation is important, big ships are more efficient.

One of the major challenges for crewed interplanetary spaceflight is the health risk presented by cosmic radiation. This is a big problem even for a 180-day journey to Mars, let alone a multi-generation trip between stars.

The linked Wiki article mentions an indicative figure of about 4 tons shielding per square metre to bring radiation levels on a space station down to roughly earthlike levels. There's a fair bit of uncertainty on that, and requirements for interstellar travel would be different again, but it gives us a ballpark: if you don't have some other way to protect against cosmic radiation, physical shielding is going to be a big part of the weight of your ship.

It's possible to have shielding that's also useful for other purposes, e.g. store the ship's water supply or hydrogen fuel in its skin. But for a generation ship, you probably want near-100% recycling of resources, meaning that the amount of water you'd otherwise need to carry is likely far less than the amount you'd need for shielding. Even if you're using consumables for shielding, you still need enough left over at the end of the trip that you're adequately shielded in the last years, which still means a large increase in the amount you have to carry.

If the weight of shielding is a major design constraint, then your ships are going to look like big balls, because that's the most efficient shape in terms of surface area per volume contained. (Edit: As mentioned in comments, some components can be put outside the shielding, so more like "big balls with stuff hanging off them"). Thanks to the square-cube law, one 200-metre-diameter ship will hold as much payload as eight 100-metre ships of similar shape, but with only half the surface area and hence half the weight of shielding.

(In fact, the big ship requires a bit less than half the shielding of eight small ships, because thickness of shielding is likely to be non-negligible, but let's not worry too much about that.)

So, if you're dependent on physical shielding for radiation protection, and you don't have some super-light unobtainium shielding, you probably want to go with a small number of big ships.

Another option might be magnetic shielding. This is a bit more speculative, but the requirements for that will probably still scale roughly with the surface area to be protected, which again makes bigger more efficient.

  • $\begingroup$ Superb answer overall. One detail: I believe the square-cube law will apply to creating magnetic fields too (a small field can be created with modest energy, but a huge field needs tons of energy), so I'm not so sure your last sentence is correct. There also might be good designs possible that leave large parts of the ship unshielded: mechanical areas where crew don't spend much time, and must wear radiation suits when they do go in for maintenance. $\endgroup$
    – Nate S.
    Dec 2, 2020 at 19:48
  • $\begingroup$ @NateS. I'm not certain either way on the magnetic shielding, but then AFAICT it's still an unproven technology, so difficult to establish its capabilities. And yes, you could save some weight by moving uncrewed components outside the shielding (and hopefully do all that maintenance by remote?) but shielding for the crewed bits is still likely to be a major design constraint. $\endgroup$
    – G_B
    Dec 2, 2020 at 21:39
  • $\begingroup$ @NateS., speaking as someone who does a lot of maintenance, doing it in a suit would be extremely difficult in many cases. If you think wedging yourself in a tight space to get at electrical or plumbing is hard normally, doing it in a suit will be even worse. And space suits don't really protect from radiation all that well. space.stackexchange.com/questions/2964/… $\endgroup$ Dec 2, 2020 at 21:40
  • $\begingroup$ @computercarguy oh absolutely; I didn't mean regular maintenance via space suit should be plan A. Ideally regular maintenance is handled by robots, and humans should only need to put on a suit for manual intervention when something unexpected happens. Weak protection can be okay if it's only occasional short term exposure, and if the area is able to be pressurized, they don't need to be full space suits, just radiation suits. $\endgroup$
    – Nate S.
    Dec 2, 2020 at 22:05
  • $\begingroup$ @NateS., even "just" a radiation suit can be a major problem for mobility, and a tear in it can get serious really quickly, depending on the radiation levels. Not to mention that they can get hot and humid really quickly. Even a painters suit gets that way fast. And I guess I didn't say it explicitly, but I didn't mean a space suit for the first part of my comment, and I meant that even a space suit doesn't provide all that great radiation protection. $\endgroup$ Dec 2, 2020 at 22:15

It depends where they start and what values you favor.

If you're launching off of Earth or some planetary object in these ships (like they show at the beginning of Wall-E), I'd recommend smaller ships. But I wouldn't recommend this strategy; either way, launching ships as big as these out of a planet's gravity well is extremely energy expensive. Instead, they should be built in space and then people shuttled up to them.

Now that we've got that out of the way, here are our options:

  1. Send one ship. Pros: sharing resources is cheaper, can be more fuel efficient if you use less engines/reactors/etc. and therefore less weight (ie big is cheaper), larger community which is probably good for psychological reasons. Cons: If anything breaks, everyone's toast.
  2. Send multiple ships. Pros: redundancy (ie if anything breaks there are other ships too), easier for each ship to manage their resources (ie no huge bureaucracy). Cons: more wasteful of resources, possible psychological effects with smaller communities

So, what matters more, being cautious about the risks of ship malfunction or saving resources? Also, which makes the better story? There's definitely a lot more tension when the sole ship malfunctions than when one ship of a fleet has problems. On the other hand, if conflicts break out between ships that could also be a story in the making.

So I'd say it comes down to which one you can better build a story out of. There are definitely arguments for each.

  • $\begingroup$ Thank you for sharing the pros and cons of both options, and recognising various things change each option. $\endgroup$ Dec 1, 2020 at 13:20
  • 1
    $\begingroup$ No problem. I've just realized over the years that often the options are close enough in value that it's best to pick the one that fits the story rather than worrying about which is the absolute most practical. Personally, I'd probably go for the multiple ships option, it just feels like there's so much you can do with that, but both are good. $\endgroup$ Dec 1, 2020 at 13:31

Frame Challenge

Is it more fuel/space efficient to have a fleet of generation ships (either several self-contained ships, or specialised ships), or to have one large ship for all functions (self-contained)?

Nothing's less efficient than a single giant sphere which gets destroyed by a passing asteroid.

That's why fuel and space efficiency are never the sole driving factors. Especially in space travel!!

Triple redundancy, over-construction, lots of machine shops and "blank" parts (generic units of material which get machined down to usable parts) will drive the designs. (Plural, since one design might have a fatal, unforeseen flaw which destroys all/most of the ships.)

That is why multiple self-contained ships launched at different times (for example, three different ships of three different designs -- total of nine ships -- launched at three different times) are the most efficient way to ensure that some people get to the destination.


Your enemy is the waste heat.

In space, you have exactly one method available to get rid of heat, and that is to radiate it away. All big spacecraft like the space shuttle and the ISS must include radiators to avoid overheating.

Generation ships have a power consumption that's roughly linear to the amount of people on board, and people on board grow with the volume of the ship ($\approx x^3$). The available surface area for waste heat radiation, however, only grows with the surface of the ship ($\approx x^2$). So, as you scale your generation ships up, you will run out of surface area for heat radiation eventually.

Now, you could say: Oh, well, then I just build huge radiators that reach out far into space. This will work for a while, but eventually you get the problem that you need to transport the waste heat from the core of the ship all the way to the tips of the heat radiators. The longer this trip becomes the less efficient the cooling will be (more energy is expended on pumping, and it becomes harder to isolate the return pipes sufficiently from the environment as they go down into the heart of the ship).

However, you can still build a huge ship that both has enough radiative surface and can withstand destruction of its parts (= the redundancy advantage of a fleet of ships):

Your ship is basically designed as a gigantic space station. It's assembled from modules and hubs that are connected via some standard connector system. Each module is basically a long tube that has its own heat radiators attached, and is connected to one hub at each end. The hubs are designed so that they mate to six modules in a single plane, and to three modules diagonally upwards. Thus one half of the nodes form a single layer of triangles with 2/5 of the modules, the other half of hubs form a second layer with another 2/5 of the modules. These two layers are connected by the last 1/5 of the modules, using the diagonal upwards connections of the hub. This forms many more triangles which are not parallel to the two planes, and thus provide lateral stiffness to the ship.

As you may know, triangular constructions are extremely stiff and never produce any bending forces on the individual beams. That's why you see such triangular construction at each and every construction crane.

You grow this ship simply by adding modules to the edge of the double plane. As such, the effective surface of the ship grows linearly with its usable volume. The mating mechanism between modules and hubs has air locks, valves on all pipes, and electrical switches on all power lines that pass through it. This allows defective modules/hubs to be separated from the ship in any manner that might be necessary. An open connection, however, allows free exchange of whatever the different modules want/need to exchange, allowing the ship to act like a big city.

  • $\begingroup$ Even your modular structure must be flat in order to radiate enough heat. If you extend it in three dimensions you don't escape the square/cube issue, modular or not (the radiated heat will be absorbed by and heat the neighbors). But in general a very good observation. $\endgroup$ Dec 4, 2020 at 16:08
  • $\begingroup$ @Peter-ReinstateMonica That's exactly why I was talking about a double plane, and growing the ship at the edge of that double plane. $\endgroup$ Dec 4, 2020 at 16:11
  • $\begingroup$ Oh. I must admit that I misunderstood the concept. Yes, planes. Good :-). $\endgroup$ Dec 4, 2020 at 16:20
  • $\begingroup$ I like this option. If you assign living quarters and working shifts correctly, i.e. don't have your specialist personnel in the same area at the same time, and have say 3 each of the "seldom used but must have" things (like hospital/med bay) widely spaced out (maybe with some being moth-balled) and with the spares stored seperately, you have the redundancy without sacrificing efficiency. $\endgroup$
    – Gwyn
    Dec 4, 2020 at 16:43

One big ship.

The answers on this so far are all much broader than the question calls for. If the question was as broad as people seem to think (i.e. which option is better), then it would be unanswerable without more info.

However, it is very specific and addresses only two things: fuel efficiency and space efficiency. It assumes both options are possible with the engineering capabilities of the world in question and doesn't care about redundancy/safety or the psychological impact or whatever other factors are normally looked at when making this type of decision.

For space efficiency, look at duplication of essential facilities (e.g. toilet/WC/loo, recreational facilities, medical facilities, etc.) as well as outside/surface area versus internal volume of the ship (square-cube law) and how this affects efficient use of space as well as available space as function of amount of construction material used. This clearly shows that the large ship is the more space efficient option as it needs less space to house and transport the same number of people.

As for fuel efficiency, assuming no strange, as yet undiscovered effects happen to larger objects in space, then as long as your ship never takes off or lands from/on a planet, the larger ship will be more fuel efficient as well. You need fuel to accelerate to journey speed and decelerate at the end and small amounts now and then to maintain journey speeds amidst various gravitational pulls. The amount of fuel you need is based on the amount of mass you need to shift, and we've already discussed how you need less material (therefore less mass) to construct one large ship than several smaller ones, so again, the one big ship will be more fuel efficient.

Just for the record, though, and completely outside the scope of the question, I'm an engineer and a firm believer in Murphy's law, so safety factor thinking says send at least 3 ships and hope at least one makes it.

  • $\begingroup$ The argument that you need more relative space and mass for infrastructure on small ships is only true for very small ships. Once your ship is above the size where there is the maximum number of people for one set of facilities, this ratio doesn't change any more with size; both scale linearly. $\endgroup$ Dec 4, 2020 at 11:55
  • $\begingroup$ Another objection is that, true: The square/cube law is probably the central variable in the equation. But that it works both ways: Yes, the "shell" volume shrinks relatively to the usable volume; but the mass needed for structural stability under acceleration grows cubic while the strength of its support structures grows only quadratic : You cannot simply scale up a small ship. $\endgroup$ Dec 4, 2020 at 11:56
  • $\begingroup$ Sure, but some things on generation ships, like medical facilities, recreational and exercise areas and engineering/machine shop, etc. would have to be duplicated in their entirety for each ship, whereas shift shedules on the big ship means you only need one of each (med bay can just ad extra beds) $\endgroup$
    – Gwyn
    Dec 4, 2020 at 14:49
  • $\begingroup$ Sur, but that argument is only valid for 20 vs 100 people, or so. For 1,000 vs 10,000 people (which is how I understand the question) the facilities scale linearly. $\endgroup$ Dec 4, 2020 at 15:06
  • $\begingroup$ Not true. For things like toilet/WC yes, but the large facilities not so much. If you have 10000 people each on 10 different ships, each med bay would need an operating theatre, anesthesia equipment, dialysis machine, X-ray, MRI, etc. each with a backup/spare. These are not needed often, but cannot be done without -when you need it, it has to be there. On a ship with 100000 people, you wil still only need one and a spare, or maybe 2, not 10 or 20. This is the same for a lot of the other bigger things like landing shuttles as well. $\endgroup$
    – Gwyn
    Dec 4, 2020 at 15:34

A larger spacecraft will be generally more efficient, all else being equal, with diminishing returns in increased efficiency as scale increases. However, a larger spacecraft is a more difficult construction project, will take more time and resources to build, will result in less efficient utilization of shipyards and other construction infrastructure, and will be less able to make use of the latest technological advances. On a colonization program level, a convoy or series of convoys may well be more efficient, even if the individual ships are less so.

Efficiency of the spacecraft can't be your primary design criteria anyway. There's always going to be a choice between launching and doing something to make the ship slightly more efficient. If you always choose to make the ship more efficient, the result is a generation ship that never leaves, but just sits in orbit being upgraded and added to forever...or more likely, gets canceled before completion.



Redundancy aka strength in numbers, does only hold up against external threads. Meaning a asteroid puncturing your ships can be protected against with lots of little turtle eggs marching to the sea. A engine-design flaw, can not be protected against with systems that all have the same flaw.To small a resource can not be protected against that way. The stranded predecessors can be scavenged though, and thus build "a road in the sky".

Redudancy is also possible within a ship, by modular design and self-recover ability. (The traders in the A-Deepness-in-the-sky universe did this). So bulkheads and decentralized command& controlposts.

For a generationship, the most important thing though, would be the capability to self-repair and bootstrap the environment up again in case of catastrophic failure. Like send everyone into storage, while it grazes for raw-material and repairs its eco-system.

  • $\begingroup$ Predecessors won't be "stranded" exactly, though. They'll be drifting further and further away at whatever velocity they'd reached when the engines died. To cannibalise them you'd need to go faster (to catch up) and then decelerate again (to match speeds), which is likely to be very expensive in whatever you're using for fuel. $\endgroup$
    – G_B
    Dec 2, 2020 at 21:49

One big ship would be most efficient in operation, as other answers have pointed out. That's only half the equation though.

This massive ship has to be built. It's massive facilities have to be able to handle huge volumes. Giant reactors for power, giant water recycling systems, giant engines to move it, giant everything.

Building large ships is pretty difficult. Parts get so large that they don't fit into existing manufacturing facilities. You can't just bolt smaller ones together; as the whole thing gets larger any joint becomes more and more of a potential weakness as the stresses involved grow too.

So before you cast that giant section of hull you have to build a giant metalworking facility that can produce vast amounts of liquid hot alloy and pour it into a giant mould. You need a kilometre long factory just to contain that extrusion.

This also increases the difficulty of maintaining the ships. Parts are now so big that they require special giant size machines to handle.

You can trade off these issues by having ships that accelerate more gently so they can be made of smaller parts connected together. Ships on Earth do that, the hull is made in sections. But then your journey time is extended, you need more supplies and so on.

Overall smaller, mass produced ships may well be more efficient over the entire lifecycle.

  • $\begingroup$ Any generation ship will be built in orbit where there are basically no restrictions in construction space, or even the weight of things that can be handled by hand. $\endgroup$ Dec 4, 2020 at 13:19
  • $\begingroup$ @cmaster-reinstatemonica it's rather difficult to form metal parts in a zero gravity forge, and you still need a really large forge. $\endgroup$
    – user
    Dec 4, 2020 at 14:19
  • $\begingroup$ But it's very easy to weld pieces together in space, and you can do any amount of preforming the parts on the planet before you send them up to the ship under construction. You need to send up everything anyways, so you might just as well send fully assembled modules, ready for being joined in zero G. That allows you to parallelize the production of the parts, as well as keeping the amount of construction people in space as low as possible. I mean, we've built a space station in space already (without even resorting to welding), that's not exactly rocket science... $\endgroup$ Dec 4, 2020 at 14:33
  • $\begingroup$ @cmaster-reinstatemonica it's not easy to weld big parts in space. For a start have you ever looked in to welding in zero gravity with no oxygen? And even though things have no weight they still have immense mass, so are not easy to handle. In fact they are harder to handle because you can't brace against the ground, so you get Newtonian physics. And the space stations we have sent up are tiny compared to what is needed for a large generation ship. $\endgroup$
    – user
    Dec 4, 2020 at 20:22
  • $\begingroup$ Well, you wouldn't weld using a torch, you would weld with electricity. And the missing oxygen is a big advantage for welding: All it does is turn the metals into oxides. You don't want oxides between your weld and the pieces you are welding. Here in our inconveniently oxygen rich atmosphere, a common technique is to blow some inert gas at the welding point, precisely to get the oxygen away (en.wikipedia.org/wiki/Gas_metal_arc_welding). As to the handling of big pieces: Look at how the ISS does it. There's a robotic arm attached to it, and astronauts can attach to the arm. $\endgroup$ Dec 4, 2020 at 23:10

In general, I think there are a lot of efficiencies that occur with scale, but I question the assertion that it's necessarily true that bigger will always be more efficient. Net efficiencies of scale don't always increase without limit for any given engineering constraint as scale goes up.

The classic example is the square-cubed relationship with surface area vs volume. As you scale one axis, volume increases faster than surface area. This is beneficial when you want to maximize how much stuff you can fit inside the ship, for example, but it's problematic for issues like heat dissipation, which relies on surface area to radiate the heat away. You can mitigate this using radiator arrays to stretch out the surface area, but if you also consider that the total amount of heat you'll need to dissipate will increase with the size of the ship, the total radiating area required might grow rather quickly. If you optimize for just the problem of heat dissipation and 'cost' of radiators, there is probably a sweet spot for this factor of efficiency.

Another example is power generation and power transmission. A large centralized power generator may be more efficient to produce, but now you also require longer distance transmission, which has associated inefficiencies. Here is another place where this a trade-off of considerations and the optimum is likely not toward infinity.

There's also the harder-to-quantify fact that scaling systems up will tend to make them more complicated, which may mean greater odds that they can fail (requiring additional redundancy if you want a safety margin) and likely requiring a larger amount of support equipment and support staff. Larger groups of people introduces a whole new slew of risks and challenges, as managing a large group of people over a long period of time to maintain a complex system creates its own challenges in terms of the required long-term social engineering. Smaller ships means smaller societies, which might be more stable in the long run than a large society sharing a single ship and having to coordinate at large scales over a long period of time to be successful.

In short, it's important to consider efficiencies of scale, but it's equally important to recognize the counterbalancing challenges/inefficiencies that occur as scale increases and weigh these factors when engineering a solution.


We don't know, and the question does not supply enough data

Breaking with tradition, I will answer a question which I have downvoted.

The question cannot be answered, except by an experienced generation ship designer who has access to data which the question chose not to include.

The point is that there are always limits in engineering. Designs work between a minimum size and a maximum size; outside the boundaries, different designs are needed: and the question does not say how to build a generation ship, how big is big, what are the design goals, how is the ship powered, and so on and so on.

Let's take a much more mundane example, and ask ourselves whether it's more efficient to carry containerized cargo from China to Europe using one big ship or a fleet of smaller ships.

  • A modern cargo ship at maximum fuel efficiency takes about 35 days to travel from China to western Europe, assuming it fits through the Suez canal. If it is too large for the Suez canal, travel time increases to about 50 days. If it can make use of the Russian North-East Passage, travel time decreases to about 25 days.

  • Current Suezmax limits are length 400 meters, beam 50 meters, draft 20 meters, height above water 70 meters. Guess what? The current largest container ships are designed to these limits.

  • But what if you spoke to the Egyptians and convince them to increase the limits? Well, you will soon run into another limit: the ship needs to be able to load and unload cargo in a port, and ports have limits too.

  • But what if you spoke to the port authorities in Antwerp, Rotterdam, Shenzhen and Shanghai to do the needful to increase the limits of their ports? You will run into other limits: for example, you would much prefer to deliveries from China every 10 days or so, and do not want to have to wait up to 70 days for your merchandize to arrive from China.

Overall, the current large container ships, capable of carrying about 24,000 TEU, are at the limit of what is actually possible and economically reasonable. We could design a container ship capable of carrying 50,000 TEU, but it would be of no use to anybody. We do not know how to design a container ship capable of carrying 500,000 TEU.

Coming back to the generation ship: let's say the current designs top out at ships about 10 km long with a diameter of about 2 km, giving an internal volume of about 31.5 km³. With great increase in cost, designing and building a ship with an internal volume of 100 km³ might be possible. But what if the colonisation effort needs 1,000 km³? Such a ship would exceed design limits, and sending multiple ships would be the only option.

In conclusion, we cannot say that one large ship is more or less efficient than multiple smaller ones unless we know a lot more about those ships than what the question says.

  • $\begingroup$ I think if I had all the numbers, I could make the calculations myself. Also, I don't know if I follow your cabal analogy? $\endgroup$ Dec 1, 2020 at 19:40
  • 1
    $\begingroup$ @Pureferret: "Cabal"? If you had all the numbers, you could have made the calculation; if we had all the numbers, we could have made the calculation. But you do not have all the numbers, and we do not have all the numbers: thus nobody can make the calculation. The question cannot be answered, because it lacks the data which would allow one to answer. $\endgroup$
    – AlexP
    Dec 1, 2020 at 19:45
  • 2
    $\begingroup$ I meant canal* not cabal. I dispute that this is unanswerable, because people have done a good job at answering it. Maybe this can't be answered purely on numbers, but others have also used logic and reasoning like you have. $\endgroup$ Dec 1, 2020 at 19:48
  • $\begingroup$ The presence of the question implies that with their current tech it's possible to build one big ship and have it function as a practical generation ship. $\endgroup$
    – Ash
    Dec 2, 2020 at 4:01
  • 1
    $\begingroup$ I disagree. If I were asking about having one ship versus multiple, even not knowing the geography or the political situation of planet Earth, I could say that things that need to be taken into account are how a big ship behaves in rocky weather, and whether it needs to pass through narrow passages. We can conjecture limitations for generation ships too. That just means this isn't hard-science. $\endgroup$
    – KeizerHarm
    Dec 2, 2020 at 9:03

Instead of answering a very open ende question, I ask why not both?

Take a huge block of ice, or an asteroid or comet. Embedded into the mass are several independent habitats and drive units. The asteroid serves as radioation shielding and reaction mass, so upon arrival it will be rather hollow. Two approaches to build such a thing come to mind immediately:

  • rockets in the back: Build rocket shaped ships, with drives at the bottom and habitat/produiction areas at the top. These are stuck into the backside of the comet. Upon arrival they fill their internal fuel tank, detach and brake individually to not waste reaction mass on brakeing the rest of the asteroid (resteroid?)

  • Drive modules and habitat modules are separated, the drive modules sit on the skin and slowly eat their way in to asteroid, tha habitats are more to the inside but distributed as much as possible

The idea is that the individual modules can help each other, but that each can be fairly self sufficient if the need arises. They should be as far from each other as the space allows so a catastrophic event is unlikely to affect all of them. The close procimity would mean one could essentially take a train from one module to another, so ne need to spend reaction masse.

An interstellar ship will be one part payload, a thousand parts or more reaction mass so you might as well use that mass for shielding and structural purposes.


As others have pointed out bigger is more efficient, but many ships are more redundant.

A couple of important things with regard to redundancy:

You want redundant design not just copies. What if you design a redundant fleet of ships but all have a design flaw in say all of the main sewerage valves.

There is also the problem (which can be used to create drama to drive the story) of independent evolution on each of the redundant ships. Both of culture of the crews and also of bacteria and viruses. Imagine Covid2099 slowly developing on one ship. Slowly enough that the crew develop immunity, but when they either arrive at the destination system or during a ship to ship transfer there is the risk of spreading an infection. We all know how problematic that can be...


Tl;dr: Huge ships are structurally less stable and have a worse mass ratio.

At the core of the matter lies the square-cube law. You cannot simply scale up a small model because the strength of its support structures grows only quadratic with size (it essentially depends on the area of the cross-sections of struts and walls) while the mass it is supporting grows with the cube of the size (it only depends on the volume). A ship that's twice as large as another ship will have, ceteris paribus, eight times the mass, but the strength of its struts will only be four times the smaller one. On large ships relatively more mass is needed for structural stability. At some point this will begin to outweigh the advantage of large ships regarding the lacking usability of an outer shell area, if it exists at all (the area would likely hold supplies anyway).

This assumes that there is no (anti) gravitation mechanism, and that the generation ship will accelerate at the beginning and the end of the journey, which seems unavoidable. It is also likely that it will need to spin for gravity which will create structural stress beyond many potential drives.


One big ship is more efficient.

The reason for this is that one ship requires less fuel, less crew, less power, and only one shipyard dock to make.

Basically, the concentration of resources is more efficient than making and sending, a bunch of smaller ships.


If that one big is drastically bigger than the smaller ships, then it's less efficient.

An example of this is the death star. The death star is so expensive and fuel hogging, that it would be much more efficient to send 2 superstar destroyers than one death star. The reason for this is because the death star is so much bigger than the superstar destroyer, that it requires at least 10 times the amount of power, fuel, and crew (and everything required to keep the crew alive and going).

Another But

Sending several smaller ships (like a couple of sizes under, like 5 cruisers instead of one capital ship) provides a greater tactical advantage.

Basically, one ship can't do flanking maneuvers, all of its weapons don't point at one target so it can't bring as much firepower to bear, and it's less maneuverable, and if it is taken out, that's it, you've lost the battle.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .