Space Elevator

There will not be a space elevator on earth until colonization of the solar system is well underway. The reason is that you can't have both low earth orbit satellites and a space elevator, they will eventually run into each other, and our society depends on satellites right now but not on a space elevator. However, two elevators on the moon are feasible today and should be done. I can't build one myself, but any bit of the design I say here can't be patented due to this page being public prior art.

SpaceX's Starship is expected to be able to put things into orbit at $10/pound. I see quotes of space elevators costing $0.60 per pound. Both are good enough to buy everyone a ticket to space. With that competition, I'm not convinced a space elevator is worth the investment, even if we could do it.

Basics

There's no material we can currently manufacture in bulk that is strong enough to build a space elevator. Carbon nanotubes and diamond nanothreads and boron nitride nanotubes and graphene are all theoretically strong enough, but they have not been made pure enough or long enough or in great enough quantities to verify their properties at scale, let alone in enough quantity and cheap enough to actually build an elevator.

Satellites (both low earth orbit and geostationary) are incompatible with a space elevator. There are about 4000 satellites in low earth orbit. They each do about 12 orbits per day. If they have to avoid a 50-meter window containing the elevator when passing over London's longitude, that's a 22,000km arc, so they have to avoid 2.2e-6 of it. 4000 satellites, 12 chances per day, is 10% random chance of some satellite hitting it per day. SpaceX is increasing the number of satellites rapidly. All the orbits precess, so each pass of each satellite has to be considered independently. You could make a single satellite avoid the elevator, and you could swing the elevator to avoid a single satellite. But when there are thousands of satellites, some satellites will do the wrong thing, there will be collisions. And that is just full-fledged satellites: space junk (which is unable to maneuver) would also cause severe collisions. The elevator would have to do all the maneuvering to avoid space junk. So a prerequisite for building a space elevator is to get rid of all the satellites and space junk. Geostationary satellites shouldn't run into the elevator, and wouldn't be going high speeds even if they did, but things falling from the elevator would hit geostationary satellites.

Either you have a space elevator (more likely many space elevators), or you have satellites. If you choose to have elevators, you can do surveillance of earth from sensors mounted at the appropriate height on the elevators. Such sensors will be geostationary no matter what their height is.

An elevator would be fixed to a stationary spot above the earth's equator, and extend out way past geosynchronous orbit, about 1/5 of the way to the moon. The elevator will want to either be in the plane of the moon's orbit, or it will try to oscillate in and out of it. This is tilted between 18 and 29 degrees from vertical on earth, would switch from north to south and back to north daily, and varies year to year. A counteracting force is the mount at the equator: if the elevator is perfectly in the moon's orbital plane, the mount will be pulled back and forth daily, which would try to make the elevator vertical from the equator. I don't know what the optimal balance between these two forces (the moon's pool and the mount's pull) will be. It is somewhere between vertical and 29 degrees off of vertical. And whatever it is, the elevator will be pulled sideways daily by both the moon and the mount.

Wikipedia (and most current literature and prototyptes) recommends supplying energy by aiming laser beams at the climber, or using electric wires from the base, or nuclear powered cars, or solar power. I've got a better solution: as cargo goes up, have even heavier cargo go down (mined from asteroids etc). Use the weight differential to power both movements.

If we had just a single cable, since cars have to go up and down at the same time, they can't be centered on the elevator. However, if they are not centered on the elevator, they will impart a sideways wave to the elevator as they travel, making it hard not to run into the elevator.

A solution is to use two cables (an up-cable and a down-cable, with horizontal spacers between the cables). That way up-cars on the up-cable and down-cars on the down-cable don't collide, and electricity can be transferred along the spacers from the down-cable to the up-cable. Cars would be C-shaped, so they can be centered on the cable yet they can avoid the spacers. Have cars cycle up one cable then down the other. Even the initial strand for constructing the elevator can easily be two strand, with up-cars and down-cars.

Better yet: have the elevator be a large cylindrical gridwork, tens of meters in diameter, with many tracks for redundancy sticking out likes gridwork fins. The grids would be about 1 part horizontal to 100 parts vertical and almost-vertical. To avoid imparting a sideways wave, cars should be centered on tracks, and tracks should impart the jerk from cars mostly via the almost-vertical bars rather than the horizontal ones. The horizontal ones have to be sliding spacers, rather than attached at fixed points.

Cross-sectional view of a space elevator, with its core, inner shell, outer shell, and tracks. About 30 meters across. No solid panels, it is all a grid of rods and spacers. The gray thing in the middle is the central core. The core is allowed to vibrate: there are no spacers between the core and the shells. The eight fins sticking out end in the tracks that cars go up and down on.

Cars

Regenerative braking doesn't actually require touching the tracks. Neither does elecromagnetic propulsion for climbers. If you're not touching the tracks, and you're out of the atmosphere, how fast you can go is pretty much limited by how well you can avoid accidentally rubbing against the tracks. Going faster means getting to orbit faster, plus any bit of cargo spends less total time weighing down the elevator. And passengers wouldn't need to bring lunch if it is fast enough. I don't know how much weight electromagnetic propulsion adds to the cars, but it should be more than made up for by the extra spacing between the cars because of going fast. Going faster is a Good Thing.

Extra weight for electromagnetic propulsion on the elevator is bad, though. We want to minimize propulsion weight. I don't know what's involved. The thyssenkrupp MULTI elevator, which uses electromagnetic propulsion rather than cables, has already explored this design space some.

Wind resistance is a problem in the atmosphere. This problem drops off rapidly, mainly in the first few km, and disappears entirely after 100 km. Geosychronous orbit is 36000 km up. So: have the cars connected in a train at the surface to minimize air resistance, going at 100 km/h again to reduce air resistance, then after air resistance is negligible (something like 5 km up) separate the cars and start accelerating at 1G. If cars can get up to 10000 km/h, passengers will reach geosynchronous orbit in 4 hours and will only need to bring lunch. Full speed is reached in a few minutes, so the total weight of cargo is only 1/100 what it would be if it were packed as close together as it was on earth's surface. The cargo at the earth's surface can be 20x the average cross-sectional weight of the elevator and still be 1/6th of the overall weight.

Cars don't want to physically touch their track, and we want to transfer power from down cars to up cars. That means electromagnetic braking down and electromagnetic propulsion up. The tracks the cars run on are supported by the elevator, they are not necessarily that strong themselves, though it helps minimize weight if they are their own support.

Cars would be on tracks outside of the elevator, and their escape plan would be to push off and fall, with the aid of parachutes.

Getting up-cars on, so that they are dense at 100km/h, is tricky. It would reduce throughput tremendously if up-cars had to join the elevator at 0km/h. I think this problem has already been solved with train tracks, and I know it has been solved with wheeled cars on roads, so load the up-cars independently, get them going, merge them into a single 100km/h train, then join that final track to the up-track. Stopping down-cars is the same thing but in reverse. At the space end, if cars are launched at escape velocity and beyond they won't get recycled anytime soon. Recycling them at the space end requires slowing them down to 0 and transferring them from an up-track to a down-track.

Most current proposals call for physically climbing the ribbon at 100km/h or less, instead of accelerating to high speeds once out of the atmosphere. The total weight of cars would be much greater. The throughput would be much less. You'd have to pack weeks of food for a trip to geosynchronous orbit. Touching the elevator causes wear and tear. I don't see any advantages to climbing the elevator so slowly, or actually touching the elevator.

Rods

There is no known material strong enough to make the elevator. I don't have any solution to that. But if one happens to come along ...

In order to avoid vibrations and increase survivability from failures, the elevator shouldn't be one solid cable. It should be a mesh of interlinked rods. A "Hoytether" seem close to right: a hollow tube formed by many rods, all mostly vertical, with each joint linked to two or more joints above and below. For some reason the Hoytether has secondary rods that are normally unstressed ... I would have all rods stressed all the time, but have either reduced stress per rod or increased load to take advantage of those extra rods.

Most current proposals for space elevators call for a single large ribbon, rather than a cylinder, on the grounds that it is less likely to fail all at once. My design with many rods forming the shells and core also avoids it failing all at once, it actually gives the elevator a bigger cross section than a ribbon would, plus it's easier to replace damaged parts in my design. That leaves the question of whether individual rods should be cylinders or ribbons. They should be cylinders. Proof: start with a cylinder 1 unit in diameter. Now increase its strength n times, either by making it a ribbon n units wide and 1 unit thick, or by making it a cylinder sqrt(n) units in diameter. The chance of a rod being damaged at all increases n times for the ribbon, but only sqrt(n) times for the cylinder. A rod failing all at once is fine, there are redundant rods and any failed or severely damaged rod will be quickly replaced.

The literature recommends tapering the cables. The elevator as a whole is whatever dimension is convenient for keeping the cars separated on their different tracks. Which is a uniform size and shape from top to bottom. It makes sense to taper the rods as lashing transfers tension from an ending rod to a starting rod. It makes sense to have the number of rods proportional to the tension the elevator needs to support at any point, so more rods in the middle, and fewer at the earth mount and the space terminal attached to the counterweight. But I expect to have a standard rod used throughout. There may be specialized rods, I just don't see a need for them so far.

Rods would be n times longer than the distance between adjacent vertical joints. Any connected adjacent joints would be connected by n+1 rods, with one ending at the joint and another starting and the other n-1 continuing through. The rods are under tremendous tension, and they need to transfer that tension to other rods before they end. One ending rod transferring it to another starting rod at the point they meet cannot be made strong enough. The trick is that if a lashing between rods can transfer only 1/t of the tension in a rod, then t lashings or more of those rods can transfer the tension. If that means rods have to overlap for distance d, they aren't under full tension for that distance, so if the rods are at least 10d long then at least 80% of each rod is under full tension. Strong enough lashings are feasible. It makes sense to taper the ends of individual rods, since the ends are not carrying as much tension as the middles, that way the weight of the tapered sections of two overlapping rods can be just a little more than the weight of a single untapered rod of the same length.

Some thin elastic coating around rods could reduce the effect of micrometeor strikes, by absorbing the smaller ones entirely and holding in fractured material for bigger ones. If you make the coating nonslippery, all a lashing needs to do is hold the rods together tightly. Regular ridges (like gear teeth) may help, to make slippage harder. Replacing a rod requires unlashing a section of the bundle (starting at one end of the rod), taking the rod out, laying a replacement rod in, then relashing. The unlashed region would move down the rod until the whole rod is removed and a new rod is put in place. Bigger rods need proportionally less coating, since the surface grows with the diameter but strength grows with the square of the diameter. Thinner rods can be shorter, bigger rods have to be longer, due to the tension they need to transfer. Either you produce rods at the site of the elevator mount, or they have to be short enough that you can carry them on trucks, or flexible enough that they can be wound around a spool.

The winner here is to be thin and flexible enough to go on a spool. That can allow km-long fibers instead of just rods the length of a truck. Transportation on the elevator itself isn't going to be a robot carrying a rod up. There's going to be a pulley system, where the rope is the good new rods (going up) and broken old rods (going down). You get one off by grabbing the head and reeling it onto a spool, then unspooling it at your leisure. Putting broken rods back on is the same thing in reverse. Frayed ends have to be cut off before broken rods are added on. Frayed ends can be cut up or ground up or packaged somehow so they can ride back down. What gets spooled is a set of rods (or broken rods) bundled together. The pulley system has to move at such a rate that enough healthy rods are pushed up the elevator to repair the elevator. If rods typically last 100 years, then the pulley system has to supply one fifty-millionth of the whole elevator per minute. The elevator is 100000km long, so that's 1/100th of the elevator's cross section on the pulley system moving 12km/hour.

Architecture

A nested tube would make it easy to link to 6 joints above and below rather than just 3. And nested tubes are much stiffer than single tubes. Rods can't end at joints because you can't transfer that much tension at just the tips. So instead, rods would continue through. Joints would just tie rods together, making them bend a little. Thread the rods so that the nearly-vertical connections are accomplished with as little bending of the rods as possible.

When a rod breaks, these are very stiff rods, they want to become straight. Broken rods sticking into cars would be very bad. This can be reduced by having fairly small rods with close-together joints on the outside of the elevator. The elevator should have two such tubes, an outer shell and inner shell, loosely connected. Sort of the shape of bamboo. With car tracks sticking out like fins from the outer shell. The nearly-vertical rods in the inner and outer shell gradually spiral around the whole shell.

Nearly all the tension should be in the vertical rods. Near-vertical rods need to exist to distribute load, but like lashing, they don't individually have to transfer much tension because there are so many of them over the course of the elevator.

If the near-vertical rods between one vertical and the next are attached not at consecutive joints, but n joints apart, then they will cross n other near-verticals in between, and can be woven together. This prevents broken near-vertical rods in the inner and outer shells from sticking out very far.

The near-vertical rods in the fins can also break. However, they are lashed either to the outer shell or the track at either end, so a break will try to be parallel with the outer shell (safe) or the track (would be safe if the broken end was fully lashed to the track). Worst comes to worst you shut down the track. The track itself can also break, which would also mean shutting down that track.

Meteor impacts and vibrations make the tube want to move. These have to be dampened enough that cars experience dislocations as swaying (or no movement at all) rather than jolts. The way to deal with that is to have a counterweight that can briefly take up the extra momentum and high-frequency oscillations. But we have to keep to minimum mass. This calls for a third tube inside the inner shell, the core, not hollow, that is allowed to do the movements that the shells may not do.

The core can use quite thick rods, with joints very far apart. It should be loosely attached to the inner shell, sort of like a string in a straw, so its rapid vibrations can be dampened by slow swaying of the shells, or absorbed at the anchor. If the core vibrates enough to hit the inner shell, cars will rub against the elevator and people die. (Exactly how to compensate for sudden impacts to these various tubes seems like an engineering problem we could solve today.) Since the elevator has excess power, that power can be used to push and pull on various components dynamically to quickly transfer fast oscillations from the shells and tracks to the core. A car moving at 10000 km/h covers 3 km per second, so waves with length under a km would be experienced as jolts rather than swaying.

Both the shells and the core try to be self-supporting, but the shells have a bunch of extra machinery that the core does not have, so the core will pull up and the shells will pull down, and there have to be nearly-vertical connecting rods between the inner shell and the core to transfer that pull (without necessarily transferring the side-to-side vibrations).

Since the whole structure is under great tension, sideways dislocations do not stay in place, they travel up or down the elevator. I don't know at what speed. However, waves travelling in the same direction as cars are probably not as dangerous as waves travelling in the opposite direction. Since the elevator has alternating up-tracks and down-tracks, another strategy for mitigating vibrations is to transfer them from a track in the wrong direction to a track in the right direction. I don't know how to do such a transfer.

Maintenance

There should be rod maintenance robots that can reach and replace broken rods. They would travel between the inner and outer shell so they do not block the cars, and so they do not get hit by the core. They have to be small enough to pass one another. In particular, a maintenance robot needs to be able to carry another broken maintenance robot out of the elevator. They are powered by the electricity generated by down-cars.

If it takes n simultaneous rod failures in particular rods to cause the whole elevator to fail, and maintenance replaces failed rods in 1/m the time betweem mean failures, the chance of random failure due to those rods is about m-n per time unit, where a time unit is how long it takes to repair a broken rod. However, failures come in multiple sizes. Assuming proper maintenance, elevator failure would typically come from large events that cause too many simultaneous rod failures. (This is very similar math to losing 3-replica or erasure-encoded files in data centers.)

Since the whole elevator is made of rods, and rods are redundant and just need to live up to specs, it is fine to start with rods of one material and swap them out for rods of another material later if you find you can manufacture a better material.

The Moon

Although there will not be a space elevator on the earth anytime soon, a space elevator on the moon is feasible today. Kevlar would work. Engineering design for space elevators should focus on building an elevator for the moon using current materials. It would help if the materials can be produced on the moon itself.

While the earth could have any number of elevators, the moon can only have two: one pointed toward earth and one away from earth. On the moon, the sun moves, but the earth is always in a fixed location in the sky. People would build houses on the earth-facing side with windows that frame their view of the earth. You can imagine two countries growing up around the two elevators: the one on the opposite side focused on the rest of the cosmos, never seeing earth at all. The other, the earth-facing one, focused on the earth, with the earth always in view. They would, of course, be named Yin and Yang. All very metaphoric.

While earth elevators can be run by the power of ore dropping down in down-cars, the moon is expected to be mostly exporting material. Beamed power from the moon to the elevator is probably the way to go. The moon also has no atmosphere, which makes beaming power more efficient.

All the problems of vibration and redundancy and maintenance remain, so moon elevators should also be nested cylinders with tracks on fins and a vibrating core and a fixed cross-section.


Bob Predicts the Future
Kite-Supported Railgun
Kessler Wind
Table of Contents