I told the story of how I had gotten involved with the JSC study of an artificial-gravity/nuclear-electric propulsion (AG-NEP) Mars vehicle study. I came into the study near the end (January 2003) and right before the Columbia disaster.
As near as I could tell, after Columbia happened, nobody kept working on the AG-NEP design, or even on Mars studies for that matter. If they did, I certainly didn’t know about it.
But for some reason, the whole idea kept rattling around in the back of my head. There were a few reasons that the JSC guys had given me that were compelling for AG-NEP as a Mars vehicle.
1. You solve the muscle and bone loss problem through artificial gravity. You don’t have to worry about hours of exercise or fret whether their bones will snap when they re-enter the Earth’s atmosphere. They’re going to be good and strong when they get home because you made sure that their bodies felt a normal level of gravity throughout the trip.
1a. Because you’ve solved the muscle and bone loss problem, the pressing need to fly the mission quickly is tremendously diminished. You can go to Mars and come back in the three-year time frame that is more astrodynamically “natural”, in other words, the time frame that aligns with the Earth and Mars’s movements around the Sun.
2. By using nuclear-electric propulsion, you actually have a credible propulsion system to execute a mission abort if you need to, for some reason along the way. You’re not going to get back quickly, but you can get back.
3. By using nuclear-electric propulsion, you actually have a credible story for vehicle reuse. You could refuel the vehicle and go again. Or you could go somewhere else like an asteroid. You have a lot more flexibility than in a chemical or NTR vehicle.
I liked the basic idea. Here was a vehicle that might actually be a true “spacecraft” as we like to think of them, with the ability to go and come from a variety of destinations and be reused. I imagined that this might be the kind of vehicle that would be in Captain Picard’s ready-room a few centuries from now as a little model, with him saying, “This is a model of the vehicle man used to explore the solar system in the early days.”
But there were definitely residual technical problems with the design as it stood when I was exposed to it. The biggest one had to do with getting the body-mounted electric thrusters to point in the right direction as the vehicle moved around the Sun, and the problem got so bad when you got to a spiral-in, spiral-out scenario around a planet that it was practically a no-go. It came down to the architectural decision to orient the thrusters so that they were firing in the same direction as the vehicle’s angular momentum vector (orthogonal to the rotation plane). That approach certainly solved any problem of plume impingement, but since the inertial direction of the thrust vector was going to change by >180 degrees during the transit to Mars, and by that much or more on the way back, you had to continuously move the angular momentum vector of your spacecraft around, and there was a non-trivial cost associated with doing that. During spiral-in or spiral-out the cost became prohibitive.
The other problem concerned spin up and spin down of the system. The JSC design assumed that spinup and spindown would be done by dedicated thrusters on the habitat module end of the vehicle. That meant a duplication in thrusters and tankage for a capability that you would want to utilize as little as possible.
Despite these problems, I recognized that the JSC design as it stood had also solved a great many problems, and that perhaps it represented a minimum in the design space of overall difficulty. I’m fond of saying “you have to pick your pain” when it comes to system optimization, and that the “best” system always involves residual problems. Perhaps this was as good as it got.
Or maybe it could be even better.
One day I was driving down the street in the pouring rain and a simple sequence of thoughts formed in my brain:
1. I had spent a whole bunch of time trying to figure out how to get solar panels on a MXER tether to point at the Sun while the tether rotated.
2. I had been lucky enough to meet Steve Canfield and had figured out how to use the Canfield joint to fix the problem of pointing the panels at an inertial target (the Sun) while the overall structure (the tether) rotated.
3. The basic problem that the AG-NEP vehicle faced was the need to point its electric thrusters at an inertial target (its thrust vector) while it rotated, much like the MXER tether needed to do with its solar arrays.
4. The reasons that JSC had rejected rotating machinery for the AG-NEP vehicle had to do with the difficulty of moving propellant and electric power across a rotating connection like a rotary joint or slip ring, and these were good and valid reasons.
5. The Canfield joint had no such problems because provided that propellant lines or power cables were flexible, they could transmit fluids and power across a Canfield joint.
thus…maybe a Canfield joint was the answer to the problems of the AG-NEP vehicle!
I couldn’t believe that I had known about the Canfield joint for so long and hadn’t put these utterly compatible ideas together.
If we were to use the Canfield joint on the AG-NEP vehicle, the overall geometry would change substantially. The logical location for the thrusters moved from the center of the vehicle, on a cross-brace, to the reactor end of the vehicle. This kept the high-power lines short since they didn’t have to run all the way to the middle of the vehicle to reach the engines. You could also place the propellant tanks on the reactor end of the vehicle as well.
This in turn led to several other vehicle advantages:
1. The moment-arm from the reactor module to the hab module is shortened (or alternatively the moment arm from the CM to the hab module can be lengthened) because now there is much more mass counterbalancing the hab module. The thrusters and the propellant constitute a lot of mass.
2. The truss between the reactor module and the hab module now doesn’t need any “cross-brace” on it or any other body-mounted structures. It can be a strong but simple extensible structure, like a CoilABLE boom, with nothing more than the power connection between the reactor and the hab module integrated into it.
3. The main thrusters can be used for spin up and spindown operations. By placing them on the end of the moment arm, they now have the ability to change the angular momentum of the vehicle, by simply remaining fixed relative to the vehicle during spinup and spindown. In fact, spin rate can be changed during thrusting simply by changing the fraction of the spin arc during which the thrusters fire.
4. The angular momentum vector of the vehicle doesn’t have to point along the thrust vector (like in the JSC design) but can point orthogonal to the spacecraft’s orbital plane. This means that the angular momentum vector’s direction doesn’t have to be altered during flight. This also means that spiral-in/spiral-out maneuvers at planets are no problem.
5. If you wanted to use the AG-NEP vehicle for asteroid missions, the electric thrusters might even be able to be used as “descent engines” provided some “landing gear” were provided on the habitat module.
6. Propellant could be used for additional reactor shielding during the flight.
Over the years since this realization, I’ve developed the capability to show how such a vehicle might look as it rotates. Here’s the link to a Java code that will show the vehicle rotating along with the Canfield joints. You can click and drag to rotate the view around and zoom in and out with the mouse wheel.
For reference, here’s the original set of slides from JSC describing the problem and their original design solution.
Latest posts by Kirk Sorensen (see all)
- Payload fraction derivation for vehicle with split delta-V (case #2) - April 22, 2022
- GEO Orbital Debris Mitigation Paper Excerpts - April 7, 2022
- Payload fraction derivation for vehicle with split delta-V (case #1) - March 14, 2021
It’s very cute, but I’m not sure if it’s actually any better than a normal 2-axis gimbal. In fact you need 3 actuators, rather than 2. I suppose the extra actuator allows you to extend it, but it’s unclear why that’s desirable.
Isn’t the twisting of the cables/pipes really more of a question of the angle you feed them? If you feed them in along any single rotary gimbal’s axis, then you get twisting (which is usually bad), but if you feed them in at 90 degrees to that, you get bending. So I don’t see that you would get any twisting with a conventional gimbal.
There might be some subtle advantages, but if so I’ve missed them.
But I do think the general idea of using a gimbal for this problem is a very good one.
It’s just that gimbals are really old tech, and the chances of coming up with a big win by reinventing them is quite low.
Really cute though.
The difficulties of explaining this idea without a physical model…you would see what I am talking about if you could hold the thing in your hand. A conventional two-axis gimbal is a serial mechanism, this is a parallel mechanism and is much stronger.
I don’t remember the exact number but the Canfield Joint is stronger in force transmission per mass than a standard gimbals. It also lets you go between any two points in it’s access envelope continuously without having to have any infinite joint velocity problems.
The Canfield Joint would allow the contacts (brushes) to be on the exterior of the plates, and never twist the supply lines. A standard gimbal would twist the lines, needing headphone-like connector that would allow for twisting. However, this would inevitably bind and destroy the line or connection.
But I still don’t get why/how it is more efficient than a CV joint. Are there numbers/explanation somewhere?
I would also figure that with a gimbal the torque is along the edges, resulting in direct power transmission.
“In fact, spin rate can be changed during thrusting simply by changing the fraction of the spin arc during which the thrusters fire.”
The spin rate can also be changed by varying the thrust sinusoidally.
Fifteen pivots instead of two, propellant feed through five pivots instead of two, yes I do not get the attraction of the Cranfield joint either, though it does look very cute.
A standard gimbal can be built using flexural elements, virtual centers for a standard gimbal are probably also better from a balancing of actuation forces perspective. I can not currently see what the Cranfield joint does that much simpler flexural joints can not do. It does not look to be any stronger/lighter to me – the arms are not exactly directly inline with load paths and the actuation forces (size of actuators), in some cases look to be extreme (unbalanced).
If one can make a flexural propellant feed system then I do not much see that it matters whether it is integrated into a Cranfield joint or not. Indeed, I expect it would be less constraining and more efficient to implement it separately.
Sorry I just do not get this either, it looks like a solution in search of a problem – very NASA. Maybe I just have different criteria for evaluating such engineering design.
“A standard gimbal would twist the lines, needing headphone-like connector that would allow for twisting. However, this would inevitably bind and destroy the line or connection.”
I find that that’s only true if you fail to fix them correctly, particularly in between the two gimbals. Provided they’re fixed and routed correctly, it just can’t do this. (This is done routinely in rocket gimbals see diagram 9-43 in Huzel & Huang for how this is done.)
“The difficulties of explaining this idea without a physical model…you would see what I am talking about if you could hold the thing in your hand.”
I watched the videos, I think I get it just fine. I just don’t like this particular mechanism, but it would clearly work, and is very cute.
I suppose AG-NEP explains EML-2 (and the ensuing thread An alternative lunar architecture at NASAspaceflight.com)
—I liked the basic idea. Here was a vehicle that might actually be a true “spacecraft†as we like to think of them, with the ability to go and come from a variety of destinations and be reused. I imagined that this might be the kind of vehicle that would be in Captain Picard’s ready-room a few centuries from now as a little model, with him saying, “This is a model of the vehicle man used to explore the solar system in the early days.‗
Or David Bowman 😉 AG-NEP is very 2001
—The reasons that JSC had rejected rotating machinery for the AG-NEP vehicle had to do with the difficulty of moving propellant and electric power across a rotating connection like a rotary joint or slip ring, and these were good and valid reasons—
Plus (negative) past experience with Galileo and Seasat…
Congratulations.
Very nice design, and yes replacing rotating power and fuel conduits with flexing conduits should greatly simplify the engineering issues.
Presumably fuel could be placed with the engines, so what you’d need to transmit through a rotating joint are electrical power and data.
Couldn’t power be transmitted through an inductive coupling, and the data wirelessly?
Note that with engines placed on joints at the reactor end of the craft the rotation spins up during one and slows down during the second part of the rotation period, to solve this the engines should be placed on the habitat end too, that would complicate things further
The net effect is zero.
Wouldn`t it be felt by the pasengers?
No, at 4 rpm, the exceptionally slight spinup and spindown would be felt over a few seconds. The engines simply don’t produce enough impulse over that period of time to be noticeable.
Interesting ideas. I’m strongly in favor of artificial gravity in conjunction with long-duration spaceflight. I was rather surprised to see which axis the thrust is being fired though. I have two questions:
1) What happens if one of the thrusters is irrevocably damaged. Does this work with 1 thruster or would you need to bring along backup thrusters (that seems like a good idea regardless).
2) Have the problems of thruster plume impingement been looked at? I suppose that the plumes of ion thrusters might expand much less than thermal rocket exhausts. That does look like a potential problem for the truss and the habitat module(s).
An ordinary electric motor can spin the passenger part of a spaceship using any fuel. The passenger compartment rotates in one direction and the core including fuel tanks spins in the other. With a nuclear or solar powered spaceship there is plenty of energy available. Artificial gravity only needs a slow spin so the motor from an electric car would do.
I believe the comparison to a gimbal is that with a gimbal rotating at a 90 degree angle the reorientation about the axis is very sudden and jerky, and likely to torque the thing horribly. A CV joint is much closer in functionality, but doesn’t support turning at a 90 degree angle. CV joints are superior for very high efficiency transmission with low vibration, but can’t handle as wide a range of angles or as much torqueing in random directions.
I agree with Bram that the ease of motion around the awkward 90-degree angle is what the designer was trying to accomplish with this joint. It also appears to lend itself better to a simple actuator set than does the CV joint, simplifying the design (the other thing the designer was probably aiming for). Using a gimbal-like device as the structural support here means that it has to be pointed by its own actuators instead of by an outside support, and the CV joint runs into more complications there than does the Canfield joint.
What about intermittent thrust, instead?
This seems to me much more simple and robust.
Let me to explain.
Suppose you are power supply bound, ie power is the bottleneck (that seems reasonable).
And suppose the electrical engines have a decent t/w and can be restarted instantaneously infinte times with no cost, so that you can afford to use them for a short time at any turn, just when they are oriented in the right direction.
I think you’ve understand the idea at this point, but I continue.
Two simmetrical engines can be mounted at the two tips of the boom. One near the crew habitat, the other near the reactor. They have to be at the same distance from the rotation centre.
The direction of thrust of the engines is tangential toward opposite rotation directions (but they can be oriented out of rotation plane, see below).
If the engines do identical pulses at the same time they produce a null torque for the veichle.
The power produced when the engines don’t operate is stored in a flyweel: you just need the capacity for a turn and a single engine pulse.
Of course the flyweel has not to disturb the rotation of the veichle, eg it could have the same rotation plane. It could be hosted in the reactor module (maybe mechanically connected to a stirling engine), or we can have a flyweel near each electric engine, that will minimize power transfert.
The engines can be gimbaled with a single degree of freedom, ie they can have an angle respect the rotation plane.
This way you can have any thrust vector regardless rotation vector. You control the 2 directions on rotation plane by pulse timing, and on the 3rd ortogonal direction by gimbaling.
Pulse duration is optimized to have a decent average t/w (that requires long pulses) and thrust efficiency (not too much in the wrong direction, ie short rotation arc during pulse). Maybe even 50% time on 50% off could be acceptable (with the worst orientation, maximum thrust angle error 45°, don’t ask me to integrate it this night to calculate average loss 🙂 … but it’s certainly less than 15% ).
No joints and no moving parts except engines 1d gimbaling.
BTW they seldom gimbal. Once the engines are oriented you’ve not to move them at any turn. Less operations, less failures.
The plumes stay clear from any part of the veichle for any possible orientation.
With this arrangement main engines could be very easily used to control orientation and rotation. Dynamics of this thing is complicatet, but probably is better to align rotation plane and orbit plane when spiraling around a planet, and stay ortogonal when you navigate in solar transfert orbit.
If the thrust pulses are annoying for the crew or the structure (but I don’t think so, they will be so weak..) you can mount the engines on proper shock absorbers or use 4 or more engines around the rotation circumference. With a “50% time on-50% off” pulse and 4 engines we don’t need flyweel power storage anymore.
I’m afraid we need other (small and possibly not critical) moving parts on the hub if we must have one or more high gain antennas pointed to Earth (very likely) or solar panels (Why? We’ve a nuclear reactor on board, after all). Canfield joint could be very useful for this.
I try to draw the “dual engine” version with ascii art:
^-CCCCC————–HHH——PPPPP———^
^ are the engines (exausts down), CCCCC is the crew habitat, PPPPP is the power supply/reactor, HHH is the hub service module in the rotation centre.
As you can see being the reactor more massive, it’s at a shorter distance from the hub than the hab module, but stil the engines levers are the same.
Propellant tank(s) has(have) to be in a symmetrical position to prevent rotation centre to move as we deplete propellant…maybe in the hub, maybe near the engines to save plumbing.
This 2d drawing can’t represent engines gimbaling, that’s in the 3rd dimension (toward you or toward the screen).
I’m aware there could be infinite reasons because this can’t work, but now I can’t see any of them. Or maybe it’s not a new idea.
But it looks so simple…
Pingback: SIPROC 2: The Canfield Joint « IPinSpace