3D printing rocket engines

[baldusi]

It occurs to me that we seem to keep thinking about an inkjet type of 3d printing.

Another approach would be more of an extrusion method where a wire or bar of metal is heated in a vacuum to a nearly moltant level, then extruded onto a platform, (which is also used to start the cooling process) in a continiously layered pattern, building up the particular part that one is trying to print.  Using a nearly moltant metal, extruded like toothpaste in a vacuum allows the metal to not only adhear to itself, but to also do so without contamination.  As this is not being reduced to a vapor, or being used in a sintering technique, the energy costs should be far lower than most other 3d metal printing techniques.

Heat transfer would be the critical issue. The cold plate would need some serious cooling loop. And the speed would be constrained by the cooling of the metal. And you'd need a very efficient airlock. Each vacuum cycling is pretty expensive, as I understand it. Both in actual gas and wear and tear.

[JasonAW3]

It occurs to me that we seem to keep thinking about an inkjet type of 3d printing.

Another approach would be more of an extrusion method where a wire or bar of metal is heated in a vacuum to a nearly moltant level, then extruded onto a platform, (which is also used to start the cooling process) in a continiously layered pattern, building up the particular part that one is trying to print.  Using a nearly moltant metal, extruded like toothpaste in a vacuum allows the metal to not only adhear to itself, but to also do so without contamination.  As this is not being reduced to a vapor, or being used in a sintering technique, the energy costs should be fa6r lower than most other 3d metal printing techniques.

Heat transfer would be the critical issue. The cold plate would need some serious cooling loop. And the speed would be constrained by the cooling of the metal. And you'd need a very efficient airlock. Each vacuum cycling is pretty expensive, as I understand it. Both in actual gas and wear and tear.

Well, you could put it trailing behind a plate outside of the ISS, to improve the vacuum and actually do the printing outside.

While we're not talking a perfect vacuum, it would be significantly better than any we could generate on Earth.

[docmordrid]

Although it's only using aluminum now (perhaps useful for the ORBITEC Vortex engine) Liquid Metal Jet Printing is pretty close to 3D extrusion, which BTW is already done with melted polymers.

Design News Blog post about LMJP....

[baldusi]

Interesting. Microgravity has the advantage on making cantilever features. But at the same time, on direct deposition methods, you sort of depend on gravity to keep it smoothly over a layer. And un even PLA surface is no problem to a 140C hot metallic print head. But metal to ceramics is a different matter. If you didn't left a smooth surface, once you try the next pass your head might interfere with excess material. And it's very difficult to assure flow will make a practically continuous surface.
Then you have the thermal issues. Air does carry heat away. Once in vacuum your only cooling methods are radiation, which is excruciatingly slow, or implementing a cold plate into the printer bed. In which case your cooling is a gradient in the z axis. What are the consequences for material deposition? I'm pretty sure a lot of thought will have to go into the G-Code regarding metal cooling, smoothing and plasticity.

[JasonAW3]

Interesting. Microgravity has the advantage on making cantilever features. But at the same time, on direct deposition methods, you sort of depend on gravity to keep it smoothly over a layer. And un even PLA surface is no problem to a 140C hot metallic print head. But metal to ceramics is a different matter. If you didn't left a smooth surface, once you try the next pass your head might interfere with excess material. And it's very difficult to assure flow will make a practically continuous surface.
Then you have the thermal issues. Air does carry heat away. Once in vacuum your only cooling methods are radiation, which is excruciatingly slow, or implementing a cold plate into the printer bed. In which case your cooling is a gradient in the z axis. What are the consequences for material deposition? I'm pretty sure a lot of thought will have to go into the G-Code regarding metal cooling, smoothing and plasticity.

The thermal issue is pretty simple as the base plate that the part is extruded to is also used as a coolant panel.  The heat would travel down (relative to the base plate) to the coolest part of the extrusion.  As the part will be in a near moltant state, outer smoothing should be pretty much as simple as having a smoothing tool traveling along behind the print head, spreading the metal into the groves created in the extrusion process.

It occurs to me that this COULD have the advantage of being able to first print the outside of a bell nozzle, allowing a web of cooling channels to be placed within the bell, and then, another print run could be done, sealing the cooling piping,(or fuel preheating piping, as you will) within the structure as an inner layer of metal is printed, bonded and smoothed to both the piping and outer bell, simultainiously.

This is an example of a potential multipass metal extrusion 3d print job that could, possibly, be done in orbit.

[baldusi]

You're forgetting to account for thermal expansion. Once you cool it contracts and stops being "stick" to the cool plate. And the contraction might curve the piece, separating the center of the plate, for example.

[JasonAW3]

You're forgetting to account for thermal expansion. Once you cool it contracts and stops being "stick" to the cool plate. And the contraction might curve the piece, separating the center of the plate, for example.

Not necessesarily.  The plate itself helps to control the cooling process allowing a regulated cooling process thus controlling heat differentials.  Attachment to the plate could be controlled by using a set of magnetic clamps that could hold the lower part of the item being fabricated against the plate.  The clamps would likewise be used to control heat flow to allow even cooling.parts that would need "quenching" as part of the initial heat treatment would require a specialized chamber into which the item would be placed, and the chamber would be partially flooded with whatever quenching solution is required.  As the solution would be sprayed directly onto and into the particular part, repressurization of the quenching chamber would not only not be required, but would in fact be inadvisable, as the solution would suddenly turn to vapor upon contact with the hot metal.  This would not only allow for the quenching process, but allow for the even dissipation of heat from the item.

[Robotbeat]

Nanfang Ventilator Co. of China has one that can print a diameter of 2.1 x 6.0 meters up to a mass of 300 tonnes. Mainly steels, including stainless, but it seems size is becoming less of a factor almost weekly.

That's not entirely accurate. The Nanfang machine is not a net-shape process, only NEAR-net shape. You still have to do a heck of a lot of post-process machining. Its only real advantages are you have a little less material wastage and a little less tool wear. It's no different than the electron beam free form fabrication that NASA Langley (which kind of pioneered the process) has been doing since at least 2000:


I mean, it has its uses, but it's nothing like the sort of 3d printing that, for instance, SpaceX is using for their SuperDracos (of course, SpaceX is merely using the process, not developing it). It can't do intricate parts (without extensive post-machining), just a basic blobby outline. However, you CAN do it in microgravity.

Still cool to watch, but the Chinese machine is by no means ground-breaking. Net-shape metal 3d printing is still incredibly limited in size (40cm cubed is the biggest machine that EOS has). But, of course, it's still useful.

All the net-shape metal 3d printers are pretty much from a bunch of German guys who used to be friends.

3d printing has a LOOOOOOOOOOOOTTTT of hype surrounding it. It is a very, very powerful tool, but a lot of the hype is kind of dumb. For instance, it will never compete in mass-production in the same way as traditional manufacturing. The per-part build time is several orders of magnitude greater. But for aerospace parts, that's just fine since you're doing relatively low-volume. And in a research environment, the technology is WAY faster than traditional manufacturing for making crude prototypes. Since you don't have to set up the tooling separately each time, you can go from idea in your head to part in your hands in an afternoon. For traditional manufacturing, that could take weeks. This allows you to iterate much, much faster with new ideas for designs. Time is very, very expensive, so in a research and development environment, this can make a HUGE difference.

Another thing: Generally, traditional type manufacturing can make you a part with much better material properties. Tolerances are way better, strength is higher, defects are lower, and stuff like surface finish and porosity is way lower for traditional manufacturing techniques. 3d printing is not usually a way to make a better, higher performance part, but to get a newly designed part made WAY faster and easier and with less fuss (and oftentimes, less red tape if you're at a place like NASA). Changes can be made very quickly, and you're generally less geometrically constrained than you are for traditional manufacturing.

[luinil]

For instance, it will never compete in mass-production in the same way as traditional manufacturing. The per-part build time is several orders of magnitude greater.

While I mostly agree with your post, in science and technology, never is a very very long time.

[Robotbeat]

For instance, it will never compete in mass-production in the same way as traditional manufacturing. The per-part build time is several orders of magnitude greater.

While I mostly agree with your post, in science and technology, never is a very very long time.

"In the same way" is also a cop-out. And granted, if you're talking 10^10000000 years or so, then yeah, maybe at that point.

But fundamentally, this statement is true. Tradition manufacturing set up in a factory line has unbeatable per-part speeds. I mean, you can have stamped parts flying out in just a few seconds while a good print of a functionally similar part would take hours. That's 3 and a half orders of magnitude different.

[KelvinZero]

Just wait until your 3D printers start printing more 3D printers out of moon dust.. Then when the moon is one big clanking replicator we will see who builds stuff faster.

[Robotbeat]

Living things are basically self replicators as you describe, but 9 women can't make a baby in 1 month.

[baldusi]

It's important to understand that Robobeat has correctly stated about single piece output per machine. Scalability, that's a whole different issue. Would 10,000 3D printers cost (acquisition and operation) than a big traditional factory? Today most certainly, in the future, probably on a case by cases basis.

[RDMM2081]

I just happened to run across this article describing a new class of 3d printed metamaterials being developed at MIT:
http://newsoffice.mit.edu/2014/new-ultrastiff-ultralight-material-developed-0619

It doesn't sound like the specific materials they are working with in this research would have any application in rocket engines, but it sounds like the process (projection microstereolithography) holds a lot of promise.  It's also a 3d printing process I'm not familiar with, does anyone know if it could be adapted to other materials which may be more applicable to rocket engines? (I.E. any metals)

[Prober]

It's important to understand that Robobeat has correctly stated about single piece output per machine. Scalability, that's a whole different issue. Would 10,000 3D printers cost (acquisition and operation) than a big traditional factory? Today most certainly, in the future, probably on a case by cases basis.

your talking about a "bot farm"

 http://forum.nasaspaceflight.com/index.php?topic=33141.msg1179912#msg1179912

Many firms have started this way (dirt cheap). 

Some firms use a finished printer to be sold to print out several parts for the next printer being built.  Many ways this is being done.

[guckyfan]

Living things are basically self replicators as you describe, but 9 women can't make a baby in 1 month.

Yes but a printer with 9 printheads may make a baby in 1 month instead of nine, depending on the geometry of the baby.

Say the baby has the shape of a rocket nozzle. 9 printheads can work at the diameter of the nozzle for most of the time until towards the top the available space cannot accomodate that number of printheads.

[KelvinZero]

And I thought I was getting hypothetical

[baldusi]

Current 3D printing methods for rocket parts use lasers and layers o nano particles. You'd need extra laser beams. But the limiting factor is more layer deposition than the sinthering part. While direct deposition is nice, it's the layers that actually allow you to do any form. With direct deposition you're limited to mostly concave when looked from below.

[JasonAW3]

Living things are basically self replicators as you describe, but 9 women can't make a baby in 1 month.

Yes but a printer with 9 printheads may make a baby in 1 month instead of nine, depending on the geometry of the baby.

Say the baby has the shape of a rocket nozzle. 9 printheads can work at the diameter of the nozzle for most of the time until towards the top the available space cannot accomodate that number of printheads.

I prefer to old fashioned way of making a baby myself.  More of a hands on operation anyway.

[john smith 19]

One of the scenarios in KE Drexlers "Engines of Creation" was construction of a rocket engine using nano assemblers.

The assemblers link together, construct "dumb" materials atom by atom then gradually "flush out" leaving honeycomb walls.

His point was that nanoassembly (like 3D printing) is additive, it excels at creating voids,  either holes or pockets inside structures.

Those FEA stress plots that show how you could lighten the structure (if you could afford the time to machine out the metal)? No problem. Don't deposit it in the first place.

I'd dispute that such systems have worse inclusions or porosity than conventionally cast or machined objects but I would say the actual crystalline quality may well be poorer. That may (or may not) be a problem.

IRL the question is more in the nature is the quality good enough to get the job done, be it prototype, or short production run (or F1 gear box gears, which were being laser depo'd in 2004).