I've got an O'Neill Cylinder that's virtually identical to Gerard O'Neill's original design, only scaled up a bit to allow for more surface area. The cylinder is divided into 6 sections, 3 windows and 3 "valleys," long stretches of habitable surface area. Mirrors reflect sunlight into the habitat, so each window directly lights one valley. The mirrors rotate to simulate a day-night cycle roughly the same as Earth's. The central axis experiences no gravity, and the atmospheric pressure there is roughly 0.8 atm.

The idea is for the cylinder to be an ecosystem as self-sustaining as possible on the long term, with large forests and bodies of water located on the valleys. My idea for the moment is for the edges of the valleys to be essentially tall mountain chains some 8 km tall, so as to keep the biomass in the valleys from "spilling" into the windows (I just assumed that would happen without some sort of barrier). General type of environment I'm going for is that of a tropical rainforest, similar to the Amazon, so high temperatures, high humidity and no seasonal variation (since the cylinder has no tilt like Earth does, that'd have to be simulated anyway).

So basically I'm scratching my head trying to figure out how the climate would work. I'm fairly certain the alternated layout of land and windows, plus the zero-g centre with slightly lower pressure, would have at least some consequence, but I don't know nearly enough physics or metereology to work that out. Would it rain on the windows? Would that disturb the water cycle or the illumination? How would air circulate? These are all smaller questions I've been having that relate to the general problem.

I also understand there's no way to answer this with any certainty, as no one's ever built this thing to test it -- all I need are some points to consider, and possible solutions to any problems that might arise from such a configuration.

Also, I'd like as little artificial climate control as possible. If things are different from Earth, but wouldn't doom life inside, then it's actually a good thing -- I like a bit of uniqueness or weirdness as long as it's harmless.

I've scoured this site and others for some sort of guidance but it seems the windowed cylinder really has gone out of fashion, because all I've found assumes a closed, windowless design with artificial lighting. Yes, I know it's better for radiation shielding, but I'm not giving up windows -- they're pretty and fit the vibe of my setting better. Thanks in advance.

  • $\begingroup$ I think that the bigger problem would be that in order to get an Earth-like gravity and a 24-hour day through rotational gravity, you'd need a structure stronger than anything available using any reasonable material could handle - something like the Orbitals of Iain Banks' The Culture series. $\endgroup$
    – nick012000
    Oct 15, 2021 at 9:31
  • $\begingroup$ @nick012000 I'm not really under the impression that such a structure can be built in reality. Honestly, I don't really think any sort of sci-fi megastructure design is realistic at all. I'm willing to handwave the materials problem for now (I assume the bigger problem you mentioned are the windows, right? I thought the hull wasn't that big a deal, since my cylinder is way smaller than the maximum-size McKendree cylinder). What I want to know is if there'd be any interesting climatic phenomenon, assuming the window-valley layout described above could be structurally stable. $\endgroup$
    – Henricoide
    Oct 15, 2021 at 13:22
  • $\begingroup$ @nick012000 point the habitat's axis the sun, in a heliocentric orbit. Place mirrors on the outside, open and close them on a 24 hour cycle. Sunset and sunrise won't be much like they are on Earth, but the day length will work OK. $\endgroup$ Oct 15, 2021 at 15:03
  • $\begingroup$ @StarfishPrime Yes, mirrors work, but then you're not deriving your day-night cycle from rotation anymore. $\endgroup$
    – nick012000
    Oct 16, 2021 at 1:40
  • $\begingroup$ @nick012000 I don't know where you got the idea that I was deriving the day-night cycle from the habitat's rotation. The mirrors rotate at different angles until they face away from the windows, it's not the entire structure that rotates. It's just like O'Neill's original design. $\endgroup$
    – Henricoide
    Oct 16, 2021 at 11:24

1 Answer 1


Unfortunately the answer here is going to be something like "whatever you want it to be".

There are a number of important outstanding questions about big spinning space habitats... seemingly simple things like "how do you work out the air pressure at a given altitude" or "does the air mass rotate smoothly in a steady state, or does it break up into enormous horizontal death tornadoes more or less immediately" inevitably gain a whole host of contradictory answers where everyone is Very Certain that their understanding of the problem is the right one. Notably, none of them have backed up their suppositions with a nice CFD model of Island Three or whatever so there's little way of telling who might be right. Given the cost of good CFD software and the technical expertise needed to operate it well, this state of affairs probably won't end any time soon.

As such, your fictional setting has two possibilities:

  • O'Neill cylinders have intractable atmospheric issues and as such can't be built. Other designs with roofs and azimuthal barriers to prevent air flow issues might be required instead. This is kinda disappointing, as you lose all of the nice features of the design, but it is the safest choice in a plausible science sort of way.
  • O'Neill cylinders work just fine. Some landscaping may be needed to control local air currents, but careful heating, cooling and humidity control ensure that the conditions inside remain human-comfortable with a range of weather conditions that can be engineered... either at will, or with some longer-term modification of the structure.

the edges of the valleys to be essentially tall mountain chains some 8 km tall

That's might also help break up axial vortices, which is a nice bonus.

Would it rain on the windows?

Not necessarily. Solar radiation will heat the land strips, causing air above them to rise. Cooler air will be drawn in from the windowed areas to either side. As the warm air rises it will cool to some degree, which can cause rain to precipitate out.

Would that disturb the water cycle or the illumination?

If it were me, I'd have the windows covered with water anyway. Makes for nice radiation shielding, and in the event of a small window puncture (say, by a micrometeorite) you'd find it easier to see atmospheric leakage. Having plenty of spare water is always a good thing. Cooling systems for the habitat might use water as the working fluid, being pumped from inside the shell of the habitat into external radiators, cooled and cycled back in, for example.

Your mountain ranges would make it hard for water to cycle back over the land areas, or for it to return to the window-seas. If the barriers were much more modest, cool air drawn over and open water surface would pick up moisture, which would then (hopefully) rain out as it blew inland and rose up. The window-lakes would then be replenished by river flow, rather than having the rivers confined to the land-slices.

This would remove any concerns about losing water, but it would bring in entirely new concerns about silting or biofouling blocking the windows. A bit of careful ecosystem design, combined with a bit of periodic dredging and suitable grazing species should help keep things clean, though.

A similar mechanism that fits your current design would have other mechanisms for letting water flow back to the window-lakes, suitably filtered to remove nutrients and so reduce the problem of fouling. This could be a passive process built into to whatever your mountain ranges are made of (probably not 8km of solid rock!)

Also, I'd like as little artificial climate control as possible.

You're probably out of luck on this one. Everything about the setup will be artificial, and the scale that it is on is simply too small to be able to soak up too much extra energy before Bad Things happen. If a forest fire started, there's the possibility that it could turn into a horizontal firestorm, and to deal with that situation you'll need a lot of very powerful emergency systems, and to avoid it happening in the first place you'll need a lot of continuous monitoring and "nudging" of weather patterns.

If you want a hands-off world, you'll need to make something bigger, in a more benign environment. A huge greenhouse on a moon or planet, for example, where there are no coriolis woes and more scope for stabilising temperatures and winds and so on.

  • $\begingroup$ Thank you for the insightful answer. That's an interesting thought about water on the windows, I once thought of that but also ran into the problem of it eventually getting the windows dirty and gave it up, thinking maybe really tall mountains would help keep the water in. As far as the artificial thing goes, I added it to the question just to see exactly what I'd have to worry about. The cylinder is supposed to have an AI "spirit" that maintains it, so hopefully I can make it work. In any case, it really does look like if I want windows, a lot of handwavium will be required... $\endgroup$
    – Henricoide
    Oct 15, 2021 at 17:04
  • $\begingroup$ "If it were me, I'd have the windows covered with water anyway" I wouldn’t. It absorbs too much light; after 30m, you'd be losing 90% of the light in best-case scenarios where it wasn't contaminated with sediments or plankton/algae. $\endgroup$
    – nick012000
    Oct 16, 2021 at 1:31
  • $\begingroup$ @nick012000 doesn't have to be that thick. 10m would be plenty. There are many options for dealing with biofouling and sediments which are outside the scope of this question. You could ask a separate one if you were interested in my suggestions. But please do consider this: the OP's habitat was built by a civilisation who can make pressurised space structures with a radius exceeding 8km, and a length likely many times that. I think they can work out how to deal with dirty windows. $\endgroup$ Oct 16, 2021 at 17:58

You must log in to answer this question.

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