• Feb. 1, 2020, 11:03 a.m.

    The idea of having ad-hoc composable O'Neill cylinder colonies where individual cylinders can be added or removed doesn't actually work.

    The problem basically comes down to physics. The cylinders are huge - anywhere from the size of a large county to the size of the state of Vermont. Since they're rotating, that means huge inertial forces, and their mass is easily large enough to generate enough gravity to be problematic. The sheer forces involved mean that any collision between cylinders is guaranteed to be catastrophic, so any collection of them absolutely must ensure that no collisions occur.

    Additionally, in order to spin up the cylinders you need some sort of counterweight. The simplest solution is simply to wrap the cylinder with another, slightly larger cylinder. The outer cylinder can be used for extra radiation and meteorite shielding, and additionally serves to counteract the natural tendency of the cylinders to precessing due to the rotation. However, this also significantly increases the mass and inertia of the system even further.

    So for example, take a tetrahedron. A tetrahedron is the simplest 3D shape you can build, and uses only 6 cylinders.
    A Tetrahedron

    The only way to hold the cylinders in this shape is to apply tension at the vertices. Applying tension at the vertices pulls the cylinders together, causing them to inevitably collide. This is a guaranteed disaster.

    To solve this problem you need a configuration which holds the cylinders apart through a balance of tensile and compressive forces. In other words, a tensegrity structure. Here is a simple tensegrity structure using the same 6 cylinders:
    Tensegrity Icosahedron

    In this example the cylinders are arranged into 3 pairs of parallel units, which each pair orthogonal to the others. None of the cylinders are touching, and the cables which hold them together also keep them apart. I honestly doubt you could build a colony much larger than this, simply due to the stress and breaking forces that increase as the distance between the units grows. Any imbalances in the forces that occur tend to be amplified the less compact the structure is, so you really want to keep it as simple and compact as you can, not to mention that gravity rapidly becomes problematic the more massive the cluster becomes.

    Obviously in a tensegrity structure you can't just pack up one of the cylinders and run off with it. Nor can you just tack on a new cylinder without completely rewiring the cables that hold it together and changing the shape entirely. Even modifying the design into something that works would be non-trivial, and quite frankly something you would probably avoid at all costs with cylinders that are already spinning. The hazards involved just aren't worth it, and it's far more convenient to move individual people around than to pack up a state-sized colony and try to fidget it around.

    An advantage of this configuration is that you can run trains along each of the cables to carry goods and people back and forth between the cylinders several times a day. In the basic configuration each cylinder is connected to 4 out of 5 of the others in the cluster, but you could easily add extra cables for full connectivity. Since the whole structure is balanced by tensile and compressive forces there's no need to expend reaction mass to get from place to place and any minor force imbalances between the mass of one train and another are absorbed by the structure itself. Unlike a more naive configuration, the movement of the trains does not drag the cylinders together or push them apart over time, and the trains can stay attached to the cables the whole time so there's no risk of missing the target at the far end like you might have if the trains were merely shot out by a mass driver and then "caught" at the far end.